Kanaoka, Katsuya (Kusatsu, JP)
Yoshikawa, Tsuneo (Nara, JP)

Plaque It!

10/545756

10/21/2008

02/16/2004

Images are available in PDF form when logged in. To view PDFs, Login  or  Create Account (Free!)

Click for automatic bibliography generation

The Ritsumeikan Trust (Kyoto, JP)

700/245

700/19, 700/45, 345/157, 345/161

G05B11/01; G06F19/00; G06F3/033; G09G5/08

700/1, 700/45, 713/300, 700/28, 713/375, 700/22, 703/2, 700/19, 700/83, 700/37, 702/182, 713/340, 700/20, 700/44, 702/181, 702/180, 703/7, 700/17

5963447	Multiple-agent hybrid control architecture for intelligent real-time control of distributed nonlinear processes	October, 1999	Kohn et al.	700/49
6208497	System and method for servo control of nonlinear electromagnetic actuators	March, 2001	Seale et al.	361/160
6362737	Object Identification system with adaptive transceivers and methods of operation	March, 2002	Rodgers et al.	340/572.1
6411944	Self-organizing control system	June, 2002	Ulyanov	706/13
6559708	Virtual and backgate supply line circuit	May, 2003	Notani	327/537
6615091	Control system and method therefor	September, 2003	Birchenough et al.	700/96
6647301	Process control system with integrated safety control system	November, 2003	Sederlund et al.	700/79
6813525	Energy management system	November, 2004	Reid et al.	700/19
6850252	Intelligent electronic appliance system and method	February, 2005	Hoffberg	715/716
6868538	Object-oriented programmable controller	March, 2005	Nixon et al.	717/100
6878066	Wireless game control units	April, 2005	Leifer et al.	463/39
6983227	Virtual models of complex systems	January, 2006	Thalhammer-Reyero	703/2
7027965	Time domain passivity control of haptic interfaces	April, 2006	Hannaford et al.	703/2
7050863	Integrated model predictive control and optimization within a process control system	May, 2006	Mehta et al.	700/29
20040024750	Intelligent mechatronic control suspension system based on quantum soft computing	February, 2004	Ulyanov et al.	707/3

JP6222839	August, 1994
JP10268904	October, 1998			METHOD AND DEVICE FOR SELF-ORGANIZING FOR OPTIMALLY CONTROL NONLINEAR PHYSICAL CONTROLLED SYSTEM

Katsuya Kanaoka et al., Passivity Monitor and Software Limiter which Guarantee Asymptotic Stability of Robot Control System, Sep. 2003, IEEE, International Conference on Robotics & Automation, p. 4366-4373.
Maschke B M J et al: “Energy-based Lyapunov functions for forced Hamiltonian systems with dissipation” Decision and Control, 1998. Proceedings of the 37th IEEE Conference on Tampa, FL, USA Dec. 16-18, 1998, Piscataway, NJ, USA, IEEE, US, vol. 4, Dec. 16, 1998, pp. 3599-3604, XP010643286 ISBN: 0-7803-4394-8.
Y. Liu et al: “Distributively controlling two robots handling an object in the task space without any communication” IEEE Transactions on Automatic Control, Oct. 1996, pp. 1193-1198, XP011000179 USA.
“Control Theory of Non-Linear Mechanical Systems—A Passivity-based and Circuit-theoretic Approach”, by S. Arimoto, Oxford Science Publication, 1996.
“Bilateral Control with Energy Balance Monitoring Under Time-Varying communication Delay”, by Y. Yokokohji, T. Imaida, and T. Yoshikawa, Proceedings of the 2000 IEEE International Conference on Robotics and Automation, p. 2684-2689, 2000.
“Control of a Flexible Manipulator with Noncollocated Feedback: Time Domain Passivity Approach”, by J.H. Ryu, D.S. Kwon, and B. Hannaford, Control Problems in Robotics, Springer Tracts in Advanced Robotics, vol. 4, p. 121-134, 2003.
B. Hannaford. et al: “Time Domain Passivity Control of Haptic Interfaces”; Proceedings of the 2001 IEEE International Conference on Robotics and Automation, vol. 2, Seoul, Korea, May 21-26, 2001, pp. 1863-1869.
Y. S. Kim, et al.; “Some Practical Issues in Time Domain Passivity Control of Haptic Interfaces”; Proceedings of the 2001 IEEE/RSJ International Conference on Intelligent Robots and Systems. vol. 3, Maui, Hawaii, USA, Oct. 29-Nov. 3, 2001, pp. 1744-1750.
B. Hannaford, et al.; “Time-Domain Passivity Control of Haptic Interfaces”; IEEE Transaction on Robotics and Automation, vol. 18, No. 1, Feb. 2002, pp. 1-10.
J. Ryu, et al.; “Stable Teleoperation with Time Domain Passivity Control”; Proceedings of the 2002 IEEE International Conference on Robotics and Automation, vol. 3, Washington D.C., May 11-15, 2002. pp. 3260-3265.
J. Ryu, et al.; “Stability Guaranteed Control: Time Domain Passivity Approach”; Proceedings of the 2002 IEEE/RSJ International Conference on Intelligent Robots and Systems, vol. 3, Lausanne. Switzerland. Sep. 30-Oct. 5, 2002. pp. 2115-2121.
J. Ryu, et al; “Sampled and Continuous Time Passivity and Stability of Virtual Environments”; Proceedings of the 2003 IEEE International Conference on Robotics and Automation, vol. 1, Taipei. Taiwan, Sep. 14-19, 2003, pp. 822-827.
J. Ryu, et al.; “Time Domain Passivity control with Reference Energy Behavior”; Proceedings of the 2003 IEEE/RSJ International Conference on Intelligent Robots and Systems, vol. 3, Las Vegas. Nevada. Oct. 27-31, 2003, pp. 2932-2937.
J. Ryu, et al.; Time Domain Passivity Control for 6 Degrees of Freedom Haptic Displays; Proceedings of the 2003 IEEE/RSJ International Conference on Intelligent Robots and Systems, vol. 3, Las Vegas. Nevada. Oct. 27-31, 2003, pp. 2944-2949.

Decady, Albert

Chang, Sunray R.

Jordan and Hamburg LLP

The invention claimed is:

1. A control system comprising: a user's control system; a virtual power monitor system interconnected with said user's control system; a bidirectional connection effecting said interconnection between said user's control system and said virtual power monitor system; said user's control system including: a user's controlled object; and a user's control strategy unit which is connected to said user's controlled object and applies a user's control strategy unit input to control said user's controlled object; and said virtual power monitor system including: a conservative control strategy unit which is connected to said user's controlled object and controls said user's controlled object conservatively though said bidirectional connection to effect a stable control; said bidirectional connection being configured to effect: connection of a user's controlled object output from said user's controlled object to said conservative control strategy; and a variable connection of a conservative control strategy unit input from said conservative control strategy unit to said user's controlled object such that said an amount of said conservative control strategy unit input applied to said user's controlled object is variable; a virtual power monitor which monitors a virtual power transmitted from said conservative control strategy unit to said user's controlled system by monitoring said bidirectional connection and calculating said virtual power as a scalar product of said user's controlled object output and said conservative control strategy unit input for said user's controlled object which is output from said conservative control strategy unit into said bidirectional connection, and such that said virtual power monitor system evaluates and analyzes stability of said user's controlled object output from said user's controlled object based on said calculating of said virtual power; said bidirectional connection varying said variable connection based on said virtual power calculated by said virtual power monitor; and a virtual power limiter controlling said variable connection and connected between said user's control strategy unit and said user's controlled object, and configured to tune a gain of said user's control strategy unit input applied to said user's controlled object based upon said virtual power monitored by said virtual power monitor, such that: said virtual power limiter tunes said gain automatically, said virtual power limiter limits the outflow of said virtual power from said user's controlled system to said conservative control strategy unit, and said virtual power limiter maintains stability of said user's controlled object output from said user's controlled object.

2. The control system of claim 1 further comprising said virtual power limiter controlling said variable connection and effecting negative feedback from said user's controlled object output to said user's controlled object input applied to said user's controlled object, and tuning a second gain of said negative feedback based on said virtual power monitored by said virtual power monitor, such that: said virtual power limiter tunes said second gain automatically, said virtual power limiter limits outflow of said virtual power from said user's control system to said conservative control strategy unit, and said virtual power limiter maintains the stability of said user's controlled object output from said user's controlled object.

3. The control system of claim 1 further comprising the virtual power limiter controlling said variable connection and being connected between said conservative control strategy unit and said user's controlled object, and tuning a second gain of said conservative control strategy unit input applied to said user's controlled object based upon said virtual power monitored by said virtual power monitor, such that: said virtual power limiter tunes said second gain automatically, and said virtual power limiter maintains stability of said user's controlled object output from said user's controlled object.

4. A control system comprising: a power assist system; a virtual power monitor system interconnected with said power assist system; and a bidirectional connection effecting said interconnection between said power assist system and said virtual power monitor system; said power assist system including: a user's controlled object controllable by a human input from a human operator; and a power assist control strategy unit connected to said user's controlled object and applying a power assist control strategy unit input to said user's controlled object to effect control of the user's controlled object such that said user's controlled object operates to effect at least one of: reduction of a physical load on said human operator; and amplification of a physical power of said human operator; and said virtual power monitor system including: human interconnection means accepting said human input from said human operator and producing a human input signal for controlling said user's controlled object voluntarily though said bidirectional connection; said bidirectional connection being configured to effect: connection of a physical controlled object output from said user's controlled object to said human operator; and a variable connection of said human input signal from said human interconnection means to said user's controlled object such that an amount of said human input signal applied to said user's controlled object is variable; and a virtual power monitor which monitors a virtual power transmitted from said human operator to said user's controlled object by monitoring said bidirectional connection and calculating said virtual power as a scalar product of said physical controlled object output and said human input signal for said user's controlled object which is output from said human interconnection means into said bidirectional connection such that said virtual power monitor system evaluates and analyzes stability of said physical controlled object output from said user's controlled object based on said calculating of said virtual power; said bidirectional connection varying said variable connection based on said virtual power calculated by said virtual power monitor; and a virtual power limiter controlling said variable connection and connected between said power assist control strategy unit and said user's controlled object, and said virtual power limiter being configured to effect a gain of said power assist control strategy unit input applied to said user's controlled object based on said virtual power monitored by said virtual power monitor, such that: said virtual power limiter tunes said gain automatically, and said virtual power limiter limits outflow of said virtual power from said power assist system to said human operator so as to ensure operator safety.

5. The control system of claim 4 further comprising said virtual power limiter controlling negative feedback from said physical controlled object output and combined with said human input signal applied to said user's controlled object, and tuning a second gain of said negative feedback based on said virtual power monitored by said virtual power monitor, such that: said virtual power limiter tunes said second gain automatically, and said virtual power limiter limits outflow of said virtual power from said power assist system to said human operator so as to ensure operator safety.

TECHNICAL FIELD

The present invention relates to a common control system in a mechanical field, an electric and electronic field, a chemical field or the like. The present invention can be applied to a wide variety of control systems of, for example, robots, machine tools, automobiles, electric and electronic circuits, chemical plants and the like.

BACKGROUND ART

[Non-patent Document 1]

“Control Theory of Non-linear Mechanical Systems—A Passivity-based and Circuit-theoretic Approach”, by S. Arimoto, Oxford Science Publication, 1996.

[Non-patent Document 2]

“Bilateral Control with Energy Balance Monitoring Under Time-Varying communication Delay”, by Y. Yokokohji, T. Imaida, and T. Yoshikawa, Proceedings of the 2000 IEEE International Conference on Robotics and Automation, page 2684-2689, 2000.

[Non-patent Document 3]

“Control of a Flexible Manipulator with Noncollocated Feedback: Time Domain Passivity Approach”, by J. H. Ryu, D. S. Kwon, and B. Hannaford, Control Problems in Robotics, Springer Tracts in Advanced Robotics, Vol. 4, page 121-134, 2003.

To secure the safety of a control system, it is necessary to guarantee that “an object to be controlled is not brought to an unstable state (does not run away out of control) under any conditions”. On this account, it is common to add a hardware-like limiter (hardware limiter) to the control system.

On the other hand, passivity (for example, refer to non-patent document 1), which is the feature of a physical system, has received attention as a theoretical scheme to guarantee stability.

A system that monitors the energy balance of a specific system with attention given to passivity to limit accumulated energy to thereby secure stability is already proposed as energy balance monitor (non-patent document 2) and passivity observer/passivity controller (non-patent document 3). The conceptual diagram of these conventionally known systems is shown in FIG. 17.

By the way, evaluation criteria of the stability of a user's controlled object, which are used for determining the operation of a hardware limiter in a common control system, are the limitation of an operating range, the limitation of current and velocity for an actuator, and the like, which are rough in distinction between stability and instability. Moreover, there is also a possibility that once a limiter is activated in a system, the system is thereafter out of control and causes a secondary disaster. Even if the system does not come to cause a disaster, in many cases, it takes a great deal of time and labor to restart the system.

Further, the energy balance monitor of the non-patent document 2 secures the passivity of a transmission line between a master and a slave and the passivity observer/passivity controller of the non-patent document 3 secures the passivity of virtual space of a haptic system, respectively, but they do not guarantee the stability of a control system including an arbitrary control strategy, an arbitrary controlled object, and an arbitrary disturbance in a real world.

Still further, in the systems described in the non-patent documents 2, 3, on the assumption that the user's controlled object 3 is passive, the user's control strategy 5 is made virtually passive on the basis of the observation of energy balance to guarantee the stability of the whole system. However, if the user's controlled object 3 is not passive, the stability of the system cannot be guaranteed. Although most of the physical systems as the user's controlled object 3 are passive, the systems cannot cope with, for example, a situation where an actuator is connected in opposite polarity by mistake to operate opposite to a driving input. If not such an extreme example, the system cannot cope with also instability caused by disturbances having direct effect on the user's controlled object.

Therefore, it is an object of the present invention to provide a system configuration that can not be realized by a conventional technology, that is, keeps control performance as high as possible for a user's control system, which has its arbitrary user's controlled object controlled by an arbitrary user's control strategy and undergoes the effect of an arbitrary disturbance, and at the same time guarantees the stability of the user's controlled object.

Further, it is another object of the present invention to provide a system configuration that can evaluate the whole user's control system including all effects from the outside world and can guarantee the stability of the whole system.

Still further, it is still another object of the present invention to provide a system configuration that can guarantee the stability of a user's controlled object even if it is not assumed that the user's controlled object is passive.

DISCLOSURE OF THE INVENTION

The present inventors have found that the above-described object can be attained by forming a system configuration in which a function (passivity monitor) of monitoring the passivity of a control system to quantitatively evaluate the stability of the control system and an intelligent limiter function (software limiter) capable of suppressing only an unstable action without interrupting control at the time of detecting the unstable action are added to an existing system, and have invented the present invention.

Moreover, the present inventors have found the following facts: in place of “monitoring the passivity of the control system” to quantitatively evaluating the stability of a control system in the related art (passivity monitor), as will be described later, power (virtual power) virtually transmitted from a conservative control strategy to a user's controlled object is monitored to evaluate and analyze the stability of a user's control system (virtual power monitor), and in place of a software limiter of functioning in such a way as to prevent passivity from being lost according to the evaluation of passivity, as will be described later, a virtual power limiter is employed which tunes the user's control system by some means on the basis of the monitoring of the virtual power in the virtual power monitor to limit the outflow of the virtual power from the user's control system and thereby functions in such a way as to guarantee the stability of the user's control system, and if a virtual power limiter system including these “virtual power monitor” and “virtual power limiter” is added to the user's control system to form a system configuration, the control system that is provided with the above-described passivity monitor and is thereby provided with a function of evaluating and analyzing the stability of the object to be controlled can be further generalized as the virtual power limiter system for guaranteeing the stability of the control system and hence can be extended in the scope of application. Under this virtual power limiter system, the configuration of each control system paying attention to the above-described passivity (refer to the first embodiment to be described later) is one example in the case where the virtual power limiter system is designed on the basis of the concept of the passivity.

Moreover, according to the virtual power limiter system of the present invention, also in a power assist system for assisting the power of a human by the use of a robot, it is possible to limit power flowing out to the human from the power assist robot irrespective of the dynamic characteristics of the human being unknown and hence to secure safety to the human. A control system of the present invention capable of attaining the above object is (1) characterized in that in an existing system having a user's controlled object controlled by a user's control strategy, the control system includes: an internal control strategy, which receives a feedback signal from the user's controlled object and inputs a control input to the user's controlled object; and a passivity monitor, which monitors the passivity of a monitored system including the existing system from the input/output of the monitored system (shown in FIG. 1), and can evaluate and analyze the stability of the user's control system.

(2) In the control system of the present invention, the user's controlled object can be typified by a mechanical system including a robot, but is not limited to this. The present invention can be applied also to various kinds of system in an electric and electronic field, a chemical field, and the like.

(3) In the control system of the present invention, an input to the monitored system can be typified by force or torque and an output from the monitored system can be typified by velocity or angular velocity, but they are not limited to these. The input to the monitored system may be a physical quantity other than force or torque and the output from the monitored system may be a physical quantity other than velocity or angular velocity.

Examples of the other physical quantity are pressure and volume change in gas, water, oil, or the like; voltage input and current output; current input and voltage output; or the like. These inputs and outputs can be arbitrarily selected if they are a pair of physical quantities for which the passivity can be defined (conjugate power pair).

(4) In the control system of the present invention, a control input from the user's control strategy to the user's controlled object and a control input from the internal control strategy to the user's controlled object may be inputted to the user's controlled object by the same physical means or by different physical means (shown in FIG. 2). Examples of the different physical means can be typified by different motors in the same robot or the like.

(5) Moreover, in the control system of the present invention, the control input from the user's control strategy to the user's controlled object and the control input from the internal control strategy to the user's controlled object have the same dimension but these two control inputs may be different from each other in the dimension.

For example, in the case where the user's controlled object is a robot having six motors, the control system may be a control system such that the control input from the user's control strategy to the user's controlled object is expressed by six-dimensional vector to all six motors and that the control input from the internal control strategy to the user's controlled object is expressed by three-dimensional vector to three motors of the six motors.

(6) One condition of the internal control strategy in the present invention, as will be described later, is to be capable of guaranteeing stability to a passive system but even an internal control strategy not satisfying this definition can be employed as an internal control strategy by adding an appropriate limiter (for example, the first software limiter) to it.

Further, the present invention is such that includes a software limiter, which prevents the passivity of the monitored system from being lost (7) by tuning the quantity of negative feedback from the output of the monitored system to the input of the monitored system (the second software limiter, shown in FIG. 3) or (8) by tuning the control input from the user's control strategy to the user's controlled object (the third software limiter, shown in FIG. 4) in the monitored system according to the evaluation of passivity by the passivity monitor and thereby guarantees the stability of the output from the monitored system.

(9) Still further, the present invention is such that includes a software limiter, which tunes the control input from the internal control strategy to the user's controlled object in the monitored system according to the evaluation of passivity by the passivity monitor (the fourth software limiter, shown in FIG. 5) and thereby guarantees the stability of the output from the monitored system and keeps controlling the user's controlled object.

(10) By appropriately determining a tuning strategy by the above-mentioned software limiter, it is possible to quantitatively and arbitrarily determine the extent to which the instability of the internal control strategy is allowed, the extent to which the instability of the user's controlled object is allowed, the extent to which the passivity of the monitored system is recovered quickly, or the extent to which the internal control strategy is substituted for the user's control strategy.

(11) In addition, by applying tuning by the above-described software limiter to the gains of the quantity of negative feedback from the output of the monitored system to the input of the monitored system, the user's control strategy, or the internal control strategy, it is also possible to automatically tune at least any one of the gains of the quantity of negative feedback, the user's control strategy, or the internal control strategy in such a way that the stability of the output from the monitored system is guaranteed.

Moreover, the present invention is characterized in that a control system provided with the above-described passivity monitor and thereby provided with a function of evaluating and analyzing the stability of an object to be controlled is further generalized as a virtual power limiter system for guaranteeing the stability of the control system, thereby being extended in the scope of application.

The above-described generalized control system of the present invention is (12) characterized in that in a user's control system in which a user's controlled object is controlled by a user's control strategy, the control system includes: a conservative control strategy, which receives a feedback signal from the user's controlled object and can input a control input to the user's controlled object; and a virtual power monitor, which monitors power virtually transmitted from the conservative control strategy to the user's controlled object, and can evaluate and analyze the stability of the user's control object.

Further, the control system of the present invention is characterized in that the control system further includes a virtual power limiter, which (13) tunes the quantity of negative feedback from the output from the user's control system to an input to the user's controlled object or (14) tunes a control input from the user's control strategy to the user's controlled object in the user's control system on the basis of the monitoring of virtual power in the virtual power monitor to thereby limit the outflow of the virtual power from the user's control system to a virtual power monitor system, and thereby guarantees the stability of the output from the user's control system.

(15) Still further, the control system of the present invention is characterized in that the control system further includes a virtual power limiter, which tunes a control input from the conservative control strategy to the user's controlled object in the user's control system on the basis of the monitoring of virtual power in the virtual power monitor, and thereby guarantees the stability of the output from the user's control system and keeps controlling the user's controlled object.

(16) In addition, according to the present invention, by applying tuning by the virtual power limiter to the gain of the quantity of negative feedback from the output of the user's control system to the input to the user's controlled object, the user's control strategy, or the conservative control strategy, it is also possible to automatically tune at least one of the gains of the quantity of negative feedback, the user's control strategy, or the conservative control strategy in such a way that the stability of the output of the user's control system is guaranteed.

Next, the above-described virtual power limiter system can be applied also to a power assist system for assisting the power of a human by the use of a robot. (17) A control system of the present invention constructed at this time is characterized in that the control system includes: a power assist system (corresponding to the above-described user's control system), which has a user's controlled object controlled by an operator and includes a power assist control strategy (corresponding to the above-described user's control strategy) that receives a feedback signal from the user's controlled object and assists to input a control input from the operator to the user's controlled object; and a virtual power monitor that monitors power transmitted from the operator to the user's controlled object, and can evaluate and analyze the stability of the user's controlled object.

Further, the control system of the present invention is characterized in that the control system further includes a virtual power limiter, which (18) tunes the quantity of negative feedback from the output from the power assist system to an input to the user's controlled object or (19) tunes a control input from the power assist control strategy to the user's controlled object in the power assist system on the basis of the monitoring of virtual power in the virtual power monitor to limit the outflow of the virtual power from the power assist system to a virtual power monitor system, and thereby limits power flowing out of the power assist system to the operator to guarantee the stability of the output from the power assist system to thereby secure safety to the operator.

(20) In addition, according to the present invention, by applying tuning by the virtual power limiter to the quantity of negative feedback from the output of the power assist system to the input to the user's controlled object, or to the power assist control strategy, it is also possible to automatically tune the gain of at least one of the quantity of negative feedback and the power assist control strategy in such a way that the stability of the output of the power assist system to secure safety to an operator is guaranteed.

Hereafter, technical terms to be used for the description of the present invention will be defined.

First, the definition of “passivity” is given. Assume that an input vector u(t) and an output vector y(t) of a certain system have the same dimension (both of them may be scalars) where t=0 is the initial time and t is the present time. At this time, the following inequality is satisfied, it is said that the system satisfies passivity (or is passive). However, in the inequality (1), γ ₀ ² is a bounded positive constant and commonly expresses internal energy at the initial time t=0 of the system.

$\begin{array}{cc}{\int }_0^t{y}^T\left(\tau \right)u\left(\tau \right)d\tau \ge -{\gamma }_0^2\left({\hspace{0.17em}}^{\forall }t>0\right)& \left(1\right)\end{array}$

“Conjugate power pair” mean a pair of physical quantities for which passivity can be defined as the input/output of a system. Examples of a conjugate power pair include force and velocity, torque and angular velocity, voltage and current, and the like. In the present invention, assume that even if the system does not always satisfy passivity in reality, a pair of physical quantities for which passivity can be defined is called a conjugate power pair.

“User's controlled object” means a physical system specified by the user of the present invention. An arbitrary system can be a user's controlled object if the system has at least a conjugate power pair as a pair of input and output. However, the user's controlled object is not necessarily passive.

“User's control strategy” means a control strategy that the user of the present invention arbitrarily specifies to control the user's controlled object. The input and output between the user's control strategy and the user's controlled object are not necessarily a conjugate power pair but can be arbitrary physical quantities selected by the user.

“Existing system” means a system in which the user's control strategy inputs a control input to the user's controlled object to control the user's controlled object. An arbitrary unknown disturbance can be input to the existing system.

In this regard, assume that “existing system” does not include the constituent elements of weights of an internal control strategy, a passivity monitor, a software limiter, and the like, which will be described later.

“Monitor system” means a system which outputs a control input from the internal control strategy to the user's controlled object and to which feedback from the user's controlled object to the internal control strategy is inputted. The monitor system can be completely externally added to an arbitrary existing system.

When the present invention will be described below, assume that the monitor system includes not only an internal control strategy but also a passivity monitor.

“Monitored system” means a system to which a control input from the internal control strategy to the user's controlled object is inputted and which outputs feedback from the user's controlled object to the internal control strategy. The monitored system includes all sections not included in the monitor system (existing system, disturbance, and the like) and becomes an object whose passivity is monitored by a passivity monitor.

“Internal control strategy” means a control strategy that feeds back the output from the user's controlled object and inputs a control input to the user's controlled object to be able to guarantee the stability of the output when the user's controlled object is passive for the input and the output. Hence, the input and output between the internal control strategy and the user's controlled object needs to be a conjugate power pair. The same physical quantities or absolutely independent physical quantities may be selected for the input and output between the internal control strategy and the user's controlled object and the input and output between the user's control strategy and the user's controlled object.

“Passivity monitor” means a means for monitoring the passivity of the monitored system for a control input from the internal control strategy with the aid of the input and output between the monitor system and the monitored system.

“Software limiter” means a system constituent element that imposes some limit to at least one section of the control system to contribute to guaranteeing the stability of the output of the user's controlled object with control performance held as high as possible. Software limiters in one embodiment of the present invention to be described below include the first to fourth software limiters.

Moreover, technical terms to be used for description of the generalized present invention will be provided.

“User's control system” means a system to which a control input from a conservative control strategy to a user's controlled object is inputted and which outputs feedback from the user's controlled object to the conservative control strategy. The user's control system becomes an object to be monitored by a virtual power monitor system.

This “user's control system” is, by definition, the same as the above-mentioned “monitored system” and is a concept capable of including not only the existing system but also the weight of the virtual power limiter and the like. The user's control system can have an arbitrary unknown disturbance inputted thereto.

“Conservative control strategy” means a control strategy that can guarantee the stability of an arbitrary controlled object satisfying some condition (for example, passivity). Moreover, the conservative control strategy can feed back an output from a user's controlled object and can input (virtually connect) a control input to the user's controlled object. For example, when a virtual power limiter system is designed on the basis of passivity, the input and output between the conservative control strategy and the user's controlled object are a pair of physical quantities that can be a conjugate power pair. The conservative control strategy can be provided independently of the user's control strategy and corresponds to the above-described “internal control strategy”.

In this regard, “virtual connection (or virtually connect)” means as follows: for example, when viewed from a virtual power monitor system, as shown in FIG. 8, a control input u _csv from a conservative control strategy 104 indeed seems to be connected to a user's control system 2 but, in reality, is subjected to a weight W _csv in the user's control system 2 (inside surrounded by a broken line in FIG. 8) and the weight W _csv is usually reduced to zero (is not connected); therefore, this is called “virtual connection”.

Moreover, only when an unstable condition occurs and the conservative control strategy needs to substitute for the control function of the user's control strategy, this control input u _csv is actually connected to the user's controlled object. However, even if the control input u _csv is actually connected to the user's controlled object, unless the weight W _csv =1, power P _v monitored by a virtual power monitor 110 is different from the quantity of power actually flowing into a user's controlled object 3 . Hence, the expression of “virtual” is used also in this sense.

Therefore, the above-described “virtual connection” expresses that a control input u _csv is at least in a state where the control input u _csv can be actually inputted although it is not certain that the control input u _csv is actually inputted.

“Virtual power” means power virtually transmitted from the conservative control strategy to the user's controlled object.

“Virtual power monitor” means a means that monitors power (virtual power) transmitted from the conservative control strategy to the user's controlled object with the aid of the input and output between the virtual power monitor system and the user's control system to thereby evaluate and analyze the stability of the user's control system and corresponds to the above-described “passivity monitor”.

“Virtual power limiter” means a system constituent element that tunes the user's control system by some means on the basis of the monitoring of virtual power by the virtual power monitor to limit the outflow of the virtual power from the user control system to thereby guarantee the stability of an output from the user control system and corresponds to the above-described “software limiter”.

Moreover, a virtual power monitor system to be described next and this virtual power limiter are generically called a virtual power limiter system.

“Virtual power monitor system” means a system, which outputs a control input from the conservative control strategy to the user's controlled object and to which feedback from the user's controlled object to the conservative control strategy is inputted, and corresponds to the above-described “monitor system”. The virtual power monitor system can be so constructed as to be completely externally added to an arbitrary user's control system.

When the present invention will be described below, assume that the virtual power monitor system includes not only the conservative control strategy but also the virtual power monitor.

As described above, the present invention provides a system construction that can be so constructed as to be completely externally added to an arbitrary existing system and quantitatively evaluates the stability of an output from an arbitrary user's control system on the basis of the monitoring of virtual power by a virtual power monitor and further guarantees the stability of an output from the user's control system. According to the present invention, it is possible to enhance the performance of the user control system as much as possible or to hold its performance as high as possible within a range in which stability can be held.

The conceptual diagram of the above-described virtual power monitor system of the present invention is shown in FIG. 8. The whole of a user's control system 2 ′ is one monitored object and the conservative control strategy 104 designed independently of a user control strategy 5 is virtually connected to the input and output of the user's control system 2 ′. By monitoring virtual power passing this connection, it is possible to evaluate the whole of the user control system 2 ′ including all effects from the outside world and to guarantee the stability of the output from the user control system 2 ′. Moreover, because the virtual power limiter system of the present invention is not predicated on the passivity of the user's controlled object, it is possible to guarantee the stability of the output from the user control system 2 ′ even if the above-described actuator of opposite polarity and disturbance exist.

In addition, a system having a plurality of control strategies, like the present invention, has other advantages. The user control strategy 5 can be an aggressive control strategy pursuing only control performance. That is, even a control strategy that is not necessarily passive is allowed if it can enhance control performance. A conservative control strategy is not related to control if it is virtually connected and hence control performance does not need to be taken into account for the design of the conservative control strategy.

On the other hand, an existing system is going to secure the passivity of the user's control strategy itself and hence cannot essentially realize only more conservative control performance.

In this manner, the present invention is very useful and widely applicable and can be easily mounted. In particular, the present invention is very useful for a user's control system that is required to guarantee stability with reliability irrespective of requiring a high degree of control or a high degree of tuning of gain.

Hereafter, the present invention will be described in more detail.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 is a diagram showing one example of a control system of the present invention.

FIG. 2 is a diagram showing another example of the control system of the present invention.

FIG. 3 is a diagram showing still another example of the control system of the present invention.

FIG. 4 is a diagram showing still another example of the control system of the present invention.

FIG. 5 is a diagram showing still another example of the control system of the present invention.

FIG. 6 is a diagram showing a monitored system of the control system of the present invention.

FIG. 7 is a diagram showing a monitor system of the control system of the present invention.

FIG. 8 is a conceptual diagram of a virtual power monitor system of the present invention.

FIG. 9 is a diagram showing one example in which a virtual power limiter system of the present invention is added to a user's control system.

FIG. 10 is a diagram showing one example in which the virtual power limiter system of the present invention is added to a power assist system.

FIG. 11 is a diagram showing one embodiment of the present invention.

FIG. 12 is a diagram showing another embodiment of the present invention.

FIG. 13 is a diagram showing still another embodiment of the present invention.

FIG. 14 is a diagram showing still another embodiment of the present invention.

FIG. 15 is a diagram showing still another embodiment of the present invention.

FIG. 16 is a diagram showing one embodiment in which the virtual power limiter system of the present invention is applied to a power assist system.

FIG. 17 is a diagram showing one example of the control system of the related art.

BEST MODE FOR CARRYING OUT THE INVENTION

1. First Embodiment

Hereafter, one embodiment and its operation of the present invention in a robot control system, which is a typical example of the object to which the present invention is applied, will be described on the basis of FIG. 1 to FIG. 7. A control system 1 in accordance with the present invention is characterized mainly in that a passivity monitor and a software limiter are added to an existing system ex. First, 1) the reason why the stability of a control system can be quantitatively evaluated by a passivity monitor will be described, and then 2) the reason why the stability of an output of a monitored system can be guaranteed by a software limiter will be described.

Here, FIG. 1 to FIG. 5 show individual configurations of a passivity monitor 10 and software limiters ( 21 a, 21 b to 24 a, 24 b ). Moreover, the respective constituent elements shown in FIG. 6 and FIG. 7 construct the control system 1 of the present invention in combination. FIG. 1 shows a control system 1 in which a monitor system 11 including an internal control strategy 4 and the passivity monitor 10 of the present invention is added to the existing system ex. FIG. 2 shows a case in which two control inputs are inputted to a user's controlled object 3 by different physical means in the control system shown in FIG. 1. FIG. 3 shows a case in which a second software limiters ( 22 a, 22 b ) of the present invention are applied to the control system shown in FIG. 1. FIG. 4 shows a case in which a third software limiters ( 23 a, 23 b ) of the present invention are applied to the control system 1 shown in FIG. 1. FIG. 5 shows a case in which the third software limiters ( 23 a, 23 b ) and the fourth software limiters ( 24 a, 24 b ) of the present invention are applied to the control system 1 shown in FIG. 1. FIG. 6 and FIG. 7 are diagrams showing the construction of the whole control system 1 of the present invention in which the existing system ex is provided with an internal control strategy 4 , the passivity monitor 10 , and the first to fourth software limiters ( 21 a, 21 b to 24 a, 24 b ). The first software limiters ( 21 a, 21 b ), as shown in FIG. 7, are provided in the internal control strategy 4 . The first software limiters ( 21 a, 21 b ) include a weight tuning strategy 21 a and a weight 21 b (W _d ). The weight W _d is connected to and controlled by the weight tuning strategy 21 a. A target value input y _d and the output y of the user's controlled object 3 are inputted to the weight tuning strategy 21 a. The output y of the user's controlled object 3 is connected to the weight W _d . An output from the weight W _d is inputted as output from the internal control strategy 4 to the passivity monitor 10 and the monitored system 2 in FIG. 6 together with output from a control strategy 6 of a PI control strategy or the like. The first software limiters built in the internal control strategy 4 can be applied also to the control system 1 shown in FIG. 1 to FIG. 5. In the following description, assume that the first software limiters ( 21 a, 21 b ) are built in the internal control strategy 4 shown in FIG. 1 to FIG. 5.

[Passivity Monitor]

First, the passivity monitor of the present invention will be described on the basis of FIG. 1. The control system 1 shown in FIG. 1 is constructed of: a monitored system 2 including an existing system ex in which a user's controlled object 3 is controlled by an arbitrary user's control strategy 5 ; and a monitor system 11 including an internal control strategy 4 capable of inputting a control input to the user's controlled object 3 and a passivity monitor 10 of a means for monitoring the passivity of the monitored system with the aid of an input and an output to and from the monitored system 2 . An input S which the user determines to realize a target value y _d can be inputted from the outside to the user's control strategy 5 and arbitrary feedback can be inputted to the user's control strategy 5 from the user's controlled object 3 . An output (control input) u _usr from the user's control strategy 5 and a control input u _int from the internal control strategy 4 are inputted to the user's controlled object 3 . An output y from the user's controlled object 3 is inputted to the passivity monitor 10 and the internal control strategy 4 . In addition, the target value input y _d can be inputted to the internal control strategy 4 . The control input u _int from the internal control strategy 4 is inputted to the passivity monitor 10 and is inputted via a weight W _int to the user's controlled object 3 . In the case of only monitoring passivity, a weight 24 b (W _int ) may be a constant. Here, by setting W _int =0, it is possible to prevent the externally added monitor system 11 from having an effect on the existing system ex.

Moreover, the control input from the user's control strategy 5 to the user's controlled object 3 and the control input from the internal control strategy 4 to the user's controlled object 3 may be inputted to the user's controlled object 3 by the same physical means or by different physical means, and a case where the control input is inputted to the user's controlled object 3 by different physical means is shown in FIG. 2. This holds for the systems shown in FIG. 3 to FIG. 5 and FIG. 14.

Let's think a robot having n degrees of freedom (having n rotary motors) as the user's controlled object 3 in FIG. 1. Assume that the joint driving torque and the joint displacement of this robot are u(t) and q(t), respectively. Let's think of controlling this robot on a target trajectory q _d (t). Here, u(t), q(t), and q _d (t) are n-dimensional vectors, respectively. At this time, an input to the user's controlled object 3 is the joint driving torque u(t) and the output y(t) from the user's controlled object 3 is expressed by a joint angular velocity, as will be described below. The output y _d (t) is a target trajectory of the joint angular velocity. Moreover, in the control system of the present invention, assume that all values except for time t are bounded.

$\begin{array}{cc}y=\frac{dq}{dt},y_d=\frac{d{q}_d}{dt}& \left(2\right)\end{array}$

When the present invention is applied to an existing system ex constructed of the above-described user's controlled object 3 and a user's control strategy 5 appropriately determined therefor, there are provided the systems shown in FIG. 1 to FIG. 5, or FIG. 6 and FIG. 7. In this system, the internal control strategy 4 is defined as follows.

$\begin{array}{cc}{u}_{\mathrm{int}}=-K_{d}y-K_P{\int }_0^t\left(y-y_d\right)d\tau -W_{d}y& \left(3\right)\end{array}$

This becomes a PI control strategy with respect to y. However, when thinking about q, this equation can be considered as a PD control strategy as shown by the following equation (4).

$\begin{array}{cc}{u}_{\mathrm{int}}=-K_P\left(q-q_d\right)-K_d\frac{dq}{dt}-W_d\frac{dq}{dt}& \left(4\right)\\ =-K_P\left(q-q_d\right)-K_{d}y-W_{d}y& \left(5\right)\end{array}$

Hereafter, in order to simplify expression, the internal control strategy 4 is expressed by equation (5). Here, u _int is an n-dimensional vector showing a control input from the internal control strategy 4 . K _p is a diagonal matrix with n rows and n columns showing a proportional gain with respect to q. K _d is a diagonal matrix with n rows and n columns showing a differential gain with respect to q. W _d is a diagonal matrix with n rows and n columns showing an tuning factor 21 b (weight) of the first software limiters ( 21 a, 21 b ) of the present invention built in the internal control strategy 4 . The software limiter will be described later.

The passivity monitor provided by the present invention is such that monitors the accumulated energy of the whole monitored system 2 of an object to provide the evaluation criterion of the stability of the control system including the arbitrary user's control strategy 5 and arbitrary environment such as disturbance. In order to evaluate the stability of the control system, the present inventors have firstly found a characteristic that a passive system is made asymptotically stable by PD control and then have found that if the passivity of the monitored system 2 is satisfied, the output y of the system can be said to be asymptotically stable. Hence, by monitoring the passivity of the monitored system 2 , it is possible to provide the evaluation criterion of the stability of the system.

Here, a passivity monitor 10 in accordance with the present invention can be expressed as follows,

$\begin{array}{cc}E={\int }_0^t{y}^T\left(\tau \right)u_{\mathrm{int}}\left(\tau \right)d\tau & \left(6\right)\end{array}$
where E shows the internal accumulated energy of the monitored system 2 . From the equation (1) of defining passivity, if the following inequality is satisfied, the monitored system can be said to be passive. E ₀ is the initial internal energy of the monitored system appropriately determined by the user.
E≧−E ₀ (7)

Hereafter, let's think the physical meaning of evaluating E by equation (7) in the passivity monitor 10 of the present invention. From equations (5) and (6),

$\begin{array}{cc}\begin{array}{c}-E={\int }_0^t{y}^T{K}_P\left(q-q_d\right)d\tau +{\int }_0^t{y}^T{K}_{d}yd\tau +{\int }_0^t{y}^T{W}_{d}yd\tau \\ ={\int }_0^t{y}_e^T{K}_P{q}_{e}d\tau -{\int }_0^t{y}_d^T{K}_P{q}_{e}d\tau +{\int }_0^t{y}^T{K}_{d}yd\tau +{\int }_0^t{y}^T{W}_{d}yd\tau \end{array}& \left(8\right)\end{array}$
where q _e and y _e are defined as errors from the target values in the following manner.
q _c =q _d −q, y _o =y _d −q (9)

Here, for convenience, when the PD control with respect to q of the internal control strategy 4 is regarded as a spring damper, the potential energy of the spring of P control is expressed by the following equation.

$\begin{array}{cc}{E}_p\left(t\right)=\frac{1}{2}{q}_c^T\left(t\right)K_p{q}_e\left(t\right)& \left(10\right)\end{array}$

From equations (8) and (10), equation (7) expressing the passivity of the monitored system 2 is found to be equivalent to the following equation, where E _p0 =E _p (0).

$\begin{array}{cc}{E}_p+{\int }_0^t{y}^T{K}_{d}yd\tau \le E_{p0}+E_0+{\int }_0^t{y}_d^T\left(K_P{q}_e\right)d\tau -{\int }_0^t{y}^T{W}_{d}yd\tau & \left(11\right)\end{array}$

Here, let's think that the above-described tuning strategy 21 a of the weight W _d is determined for an appropriately determined constant E _pmax by the following inequality (12). By thinking in this manner, even if the internal control strategy 4 applied to the monitored system 2 varies with time or even if an excessive target value input y _d is inputted at some timing to the internal control strategy 4 , the first software limiter operates to increase weight W _d , thereby guaranteeing the stability of the output y of the monitored system 2 as is the case where the internal control strategy 4 does not vary with time.

$\begin{array}{cc}{E}_{p0}+E_0+{\int }_0^t{y}_d^T\left(K_P{q}_e\right)d\tau \le E_{p\max }+{\int }_0^t{y}^T{W}_{d}yd\tau & \left(12\right)\end{array}$

All variables except for time t are bounded and hence with the passage of time, the weight W _d that always satisfies the above inequality can be determined. At this time, from equations (11) and (12), the following inequality can be obtained.

$\begin{array}{cc}{E}_p+{\int }_0^t{y}^T{K}_{d}yd\tau \le E_{p\max }& \left(13\right)\end{array}$

The first term on the left side of equation (13) is potential energy accumulated in the spring of P control at time t, and the second term is the total quantity of energy dissipated by the damper of D control. When this inequality is satisfied, the energy dissipated by the damper is surely positive and monotonously increases and its upper value is limited by the constant E _pmax on the right side. Hence, it is found that the potential energy of the first term on the left side decreases asymptotically. That is, if the monitored system 2 is monitored by the passivity monitor 10 and its passivity is satisfied, the inequality (13) is equivalently satisfied, which results in showing the asymptotic stability of the joint displacement q.

Therefore, if E is computed by the passivity monitor 10 and it is evaluated that the monitored system 2 is passive, it is found that the asymptotic stability of the output y of the monitored system 2 is also guaranteed.

On the other hand, if the monitored system 2 is not passive, the stability of the output y of the monitored system 2 is not necessarily guaranteed, and it is evaluated that as the output E from the passivity monitor 10 becomes smaller, the monitored system 2 further generates energy and becomes more unstable.

Moreover, E ₀ in inequality (7), which is a threshold for determining whether the system becomes stable or unstable, is a value to be arbitrarily set by the user and is such a quantity having a clear physical meaning that is the initial internal energy of the monitored system 2 , and hence can be quantitatively evaluated. Similarly, also the constant E _pmax of weight tuning strategy, expressed by inequality (12) is such a quantity having a clear physical meaning that is the upper limit of internal energy of the monitored system 2 of PD control and hence can be quantitatively evaluated.

[Guarantee of Stability by Software Limiter]

Next, the guarantee of the stability of output of the monitored system by the software limiter will be described. The whole control system in which the first to fourth software limiters ( 21 a, 21 b to 24 a, 24 b ) are applied to the control system in FIG. 1 is shown separately in FIG. 6 and FIG. 7. Moreover, a control system in which not only the first software limiters but also the second software limiters ( 22 a, 22 b ) are applied to the control system in FIG. 1 is shown in FIG. 3; a control system in which the third software limiters ( 23 a, 23 b ) are applied to the control system in FIG. 1 is shown in FIG. 4; and a control system in which the third and the fourth software limiters ( 23 a, 23 b, 24 a, 24 b ) are applied to the control system in FIG. 1 is shown in FIG. 5, respectively.

The connection of the respective software limiters will be described on the basis of FIG. 6 and FIG. 7. The first software limiters are shown in FIG. 7 and are connected in the manner as described above. The second to fourth software limiters will be described. These also include weight tuning strategies ( 22 a to 24 a ) and weights W _v , W _usr , and W _int ( 22 b to 24 b ) which are determined and controlled by the weight tuning strategies ( 22 a to 24 a ). To each of the weight tuning strategies ( 22 a to 24 a ) is inputted the output E of the passivity monitor 10 . To the respective weights W _v , W _usr and W _int are inputted the output y from the user's controlled object 3 , the control input u _int from the user's control strategy 5 , and the control input u _int from the internal control strategy 4 . The output of each of the weights W _v , W _usr , and W _int is inputted to the user's controlled object 3 . Here, the output from the weight W _usr and the output from the weight W _int are inputted to the user's controlled object 3 by the same physical means as shown in FIG. 6 or by different physical means as shown in FIG. 3 to FIG. 5.

In the control system 1 shown in FIG. 6 and FIG. 7, the input u to the user's controlled object 3 is expressed as follows,
u=W _usr u _usr +W _int u _int −W _v y+u _dis (14)
where u _usr is an n-dimensional vector showing a control input from the user's control strategy 5 ; y is an n-dimensional vector showing an output from the monitored system 2 ; u _dis is an n-dimensional vector showing an input torque from disturbance; W _v is the tuning factor 22 b (weight) of the second software limiters ( 22 a, 22 b ) of the present invention; W _usx is the tuning factor 23 b (weight) of the third software limiters ( 23 a, 23 b ) of the present invention; and W _int is the tuning factor 24 b (weight) of the fourth software limiters ( 24 a, 24 b ) of the present invention. Each of the above W _v , W _usx , and W _int is a diagonal matrix with n rows and n columns and the tuning strategy of the weight W _int is expressed by the following equation.
W _usr +W _int =I (15)
where I is a unit matrix with n rows and n columns. At this time, the following equation is obtained from equations (6), (7), (14), and (15).

$\begin{array}{cc}{\int }_0^t{y}^T{W}_{\mathrm{usr}}\left(u_{\mathrm{usr}}-u_{\mathrm{int}}\right)d\tau +{\int }_0^t{y}^T{U}_{\mathrm{dis}}d\tau \le E_0+{\int }_0^t{y}^{T}ud\tau +{\int }_0^t{y}^T{W}_{v}yd\tau & \left(16\right)\end{array}$

This equation is equivalent to equation (7) and hence if this inequality is satisfied, the passivity of the monitored system 2 is satisfied. Now, the user's controlled object 3 is a robot and passivity holds between the input and the output for the robot, and hence the second term on the right side is limited by a lower limit as shown by inequality (1). Hence, even if a condition where equation (16) is not satisfied occurs, if a tuning strategy that makes W _usr get asymptotically close to 0 is selected as the tuning strategy of the weight W _usr and a tuning strategy that asymptotically increases W _v is selected as the tuning strategy of the weight W _v , the weights W _usr and W _v that satisfy inequality (16) surely exist with the passage of time because all values except for time t are bounded. If these W _usr and W _v are set at the weights of limiters, inequality (16) is satisfied. That is, this is equivalent to the monitored system 2 being passive. When the monitored system 2 is passive, as described above, the asymptotic stability of the output y of the monitored system 2 is guaranteed by the internal control strategy 4 . Therefore, it has been shown that the asymptotic stability of the output y of the monitored system 2 is guaranteed by the software limiters.

Here, this does not say that if a user's controlled object is not a user's controlled object like a robot in which passivity holds between the input and the output, the passivity of the monitored system is not shown, but even if a user's controlled object does not satisfy passivity, if inequality (16) can be satisfied by tuning W _usr and W _v , the passivity of the monitored system holds without problem.

By updating the internal control strategy 4 and the user's control strategy 5 in the form of including W _v , W _usr , and W _int ( 21 b to 24 b ), which are tuned in the above-described manner by the weight tuning strategies ( 21 a to 24 a ) of the software limiters, as gains, the weight tuning strategies ( 21 a to 24 a ) can be used as automatic gain tuning mechanisms capable of always securing asymptotic stability. With this, it is possible to tune a gain appropriately without depending on trial and error and without causing instability.

2. Second Embodiment

Next, a virtual power limiter system that is newly defined by generalizing a system including the above-described passivity monitor and software limiters and guarantees the stability of a control system will be described.

First, a virtual power monitor (corresponding to the passivity monitor 10 of the first embodiment), which is one constituent element of the virtual power limiter system of the present invention, will be described.

In FIG. 8 is shown a user's control system 2 ′ having a virtual power monitor 110 added thereto. Assume that a control input from a conservative control strategy 104 , which is designed separately from the user's control strategy 5 , is u _csv (t). The control input u _csv from the conservative control strategy 104 does not need to be necessarily inputted in reality to the user's controlled object 3 but is virtually connected to the user's controlled object 3 .

Here, the quantity of monitoring by the virtual power monitor 110 is defined as the following equation.
P _v ( t )= y ( t ) u _csv ( t ) (17)

This P _v is the power virtually transmitted from the conservative control strategy 104 to the user's control object 3 . However, this power is not necessarily a physical quantity expressed by [W]. The stability of the user's control system 2 ′ can be evaluated from the result of monitoring P _v of the virtual power monitor.

Moreover, a virtual power limiter (corresponding to the software limiter in the first embodiment), which is another important element constructing the virtual power limiter system of the present invention, will be described.

In FIG. 9 is shown a conceptual diagram of a virtual power limiter system 100 . A virtual power limiter 102 tunes the user's control system 2 ′ on the basis of the quantity of monitoring P _v to thereby guarantee the stability of the user's control system 2 ′. In this embodiment, the virtual power limiter 102 is constructed of respective weights 122 b, 123 b, and 124 b (W _v , W _usr , and W _csv ) and a weight tuning strategy (corresponding to a user's control system tuning strategy in the drawing) 120 .

Next, it will be shown that the virtual power limiter system 100 , which is designed as one embodiment of the present invention on the basis of the concept of passivity, can guarantee the stability of the user's control system 2 ′. Here, the passivity is defined by equation (1).

First, to evaluate the stability of the user's control system 2 ′, let's think about the following quantity computed by the quantity of monitoring P _v ,

$\begin{array}{cc}{E}_v\left(t\right)={\int }_0^t{P}_v\left(\tau \right)d\tau ={\int }_0^t{y}^T\left(\tau \right)u_{\mathrm{crv}}\left(\tau \right)d\tau & \left(18\right)\end{array}$
where E _v is energy virtually transmitted from the conservative control strategy 104 to the user's control system 2 ′. Moreover, the following inequality is introduced as the evaluation criterion of the stability of the user's control system 2 ′. Here, E _v0 is an arbitrary constant previously determined.

$\begin{array}{cc}{E}_v\left(t\right)={\int }_0^t{y}^T\left(\tau \right)u_{\mathrm{crv}}\left(\tau \right)d\tau \ge -E_{v0}& \left(19\right)\end{array}$

Here, the state where the above inequality does not hold physically means that the user's control system 2 ′ generates energy and where power excessively flows out. At this time, by adopting one of strategies of 1) increasing the internal dissipated energy or 2) decreasing the internal generated energy for the user's control system 2 ′, the user's control system 2 ′ can be tuned in the direction in which the inequality (19) is satisfied.

To realize the first strategy, it is thought that a tunable energy dissipating element is added to the inside of the user's control system 2 ′ to increase the dissipated energy. To realize the second strategy, it is thought to decrease the generated energy by making the user's control strategy 5 , which has a high possibility of being an energy generating element, a more conservative control strategy.

The virtual power limiter applies the above strategy according to the quantity of monitoring P _v to limit power flowing out of the user's control system 2 ′. If the user's control system 2 ′ can be tuned in such a way that the inequality (19) is always satisfied irrespective of whether or not the control input (manipulated variable) u _csv is virtually inputted from the conservative control strategy 104 , the whole system becomes a robust control system and has stability guaranteed.

[i) Case of Applying the Present Invention to a Robot Control System of a User's Control System]

An example of applying the virtual power limiter system 100 of the present invention to a robot control system will be described. FIG. 9 shown above is a block diagram showing its configuration example. Here, assume that the detail of the user's control strategy 5 and the dynamic characteristics of robot hardware as the user's controlled object 3 are unknown. However, assume that each joint velocity y(t) of the robot can be measured by a sensor and that a control input u _usr (t) from the user's control strategy to each corresponding joint actuator can be tuned. Here, y and u _csv are a conjugate pair.

A driving force u actually applied to the robot hardware is as follows,
u=W _usr u _usr +u _dis +W _csv +u _csv +W _v u _v (20)
where u _dis (t) is an unknown input such as disturbance, u _v (t)=−K, y(t) is a negative output feedback connected to the user's controlled object, and K _v is a positive diagonal matrix of its gain.

By the way, although equation (20) corresponds to equation (14), when compared with each other, both equations are different from each other in expression relating to the weight W _v . This is because the elements of the above-described matrix K _v of the gain are already included in the weight W _v in the expression by equation (14), whereas the elements of the matrix K _v of gain are not included in the weight W _v in the expression by equation (20), that is, the elements of the matrix K _v are separated from the weight W _v as shown by u _v (t)=−K _v y(t).

Therefore, in FIG. 3 and FIG. 6 showing the first embodiment of the present invention, only the block 22 b of the weight W _v is shown at the point of negative output feedback connected between the input and the output of the user's controlled object, whereas in FIGS. 9 to 16 showing the second embodiment of the present invention, not only the block 122 b of the weight W _v but also the block 122 c of the gain −K _v of the negative output feedback is shown at the point of negative output feedback connected between the input and the output of the user's controlled object.

Here, each of W _usr (P _v ), W _csv (P _v ) and W _v (P _v ) is a positive diagonal matrix expressing a variable weight constructing a virtual power limiter. The diagonal elements of W _usr , W _csv are not less than 0 and not more than 1 and W _csv =I−W _usr , where I is a unit matrix.

As the conservative control strategy 104 , for example, the following each shaft PD control is selected,
u _csv =−K _p ( q−q _d )− K _d y (21)
where K _p and K _d are positive diagonal matrixes and q _d is the target value command of each joint displacement and is fundamentally made to agree with a target value command to the user's control strategy.

At this time, from equation (20), the evaluation criterion (19) of stability becomes equivalent to the following inequality.

$\begin{array}{cc}{\int }_0^t{y}^T{w}_{\mathrm{usr}}\left(u_{\mathrm{usr}}-u_{\mathrm{crv}}\right)d\tau +{\int }_0^t{y}^T{u}_{\mathrm{dis}}d\tau \le E_{v0}+{\int }_0^t{y}^{T}ud\tau +{\int }_0^t{y}^T{w}_v{K}_{y}yd\tau & \left(22\right)\end{array}$

From the above inequality, it is found that when a state where inequality (19) is not satisfied is detected by monitoring P _v , by bringing W _usr close to 0 and by increasing W _v , the user's control system 2 ′ can be tuned in a such way that inequality (19) is satisfied.

If inequality (19) is always satisfied by suppressing the effect y ^T u _dis of unknown disturbance and the effect y ^T u of unknown dynamic characteristics of the robot hardware, the dissipation of output of the whole system is shown by each shaft PD control strategy (21) and the stability of the unknown robot control system can be guaranteed.

[ii) Case of Applying the Present Invention to Power Assist System]

Moreover, an example will be described in which the virtual power limiter system of the present invention is applied to a power assist system (corresponding to the above-described user's control system 2 ′) for assisting human power by the use of the same robot as described above. It is said that the conventionally known power assist system is generally hard to guarantee stability because an unknown dynamic characteristic of a human is included in a feedback loop. However, if the present invention is applied to the power assist system, in spite of the fact that the dynamic characteristics of the human is unknown, it is possible to restrict power flowing out to the human from the power assist system and hence to secure safety for the human by a virtual power limiter.

FIG. 10 is a block diagram showing one example to which the virtual power limiter system 100 of the present invention is applied to a power assist system 150 .

When the present invention is applied to the power assist system 150 , the operating force of the human shown in the following is used as the conservative control strategy 104 .
u _csv =u _h (23)

This is an operating force applied directly to a power assist robot 103 (corresponding to the above-described user's controlled object 3 ) by the human. Since the human (operator H) is connected mechanically directly to the power assist robot 103 , virtual connection is not applicable and hence the weight of the virtual power limiter W _csv =I. In this embodiment, assume that the simplest power assist control strategy, as shown in the following, is used for the user's control strategy (corresponding to 111 in FIG. 10, hereafter referred to as “power assist control strategy” in the case where the present invention is applied to the power assist system 150 ). Other higher power assist control strategy may be used.
u _usx =K _usr u _h (24)
where K _usr is a positive diagonal matrix expressing a power assist gain. By substituting these into equation (20), the driving force applied to the power assist robot 103 is expressed by the following equation.
u=u _h +W _usr K _usr u _h +u _dis +W _v u _v (25)

From this equation (25), the evaluation criterion (19) of stability becomes equivalent to the following equation.

$\begin{array}{cc}{\int }_0^t{y}^T{w}_{\mathrm{usr}}{K}_{\mathrm{usr}}{u}_{h}d\tau +{\int }_0^t{y}^T{u}_{\mathrm{dis}}d\tau \le E_{v0}+{\int }_0^t{y}^{T}ud\tau +{\int }_0^t{y}^T{w}_v{K}_{v}yd\tau & \left(26\right)\end{array}$

Therefore, it is clear that also in the case of the power assist system, by bringing the weight W _usr of the virtual power limiter close to 0 and by increasing W _v , the power assist system can be tuned so that equation (19) is satisfied. This corresponds to the realization of automatic gain tuning of preventing power assist from being unstable by the virtual power limiter system 100 .

In this manner, according to the present embodiment, it is possible to limit power flowing out to the human (operator H) from the power assist system 150 irrespective of human dynamic characteristics being unknown and hence to secure safety to the human by the virtual power limiter.

In this regard, in the respective examples of the above-described embodiments i) and ii), even if the passivity of the robot hardware ( 3 , 103 ) is not necessarily supposed, when the virtual power limiter system 100 is designed on the basis of the concept of passivity, if only inequality ( 22 ) and inequality ( 26 ) are satisfied by tuning variable weights 122 b, 123 b, and 124 b (W _v , W _usr , W _csv , here, in the case of power assist system 150 , W _csv =I), it is possible to guarantee stability also for user's controlled object which is not passive.

Here, assume that when the virtual power limiter system 100 of the present invention is applied to the user 5 control system as shown in FIG. 9, the tuning strategy of the respective weights is a user's control system tuning strategy 120 and that when the virtual power limiter system 100 of the present invention is applied to the power assist system as shown in FIG. 10, the weight tuning strategy of the respective weights is a power assist system tuning strategy 120 ′.

Embodiment 1

[Various User's Control Systems to Which the Virtual Power Limiter System is Applied]

Hereafter, one embodiment of the present invention will be described with reference to the accompanying drawings.

Here, in the descriptions of the respective following examples, the same parts as in FIG. 1 to FIG. 10 are denoted by the same reference symbols and the like.

FIRST EXAMPLE

Basic Application

FIG. 11 is a block diagram showing one example of a motor control system to which the virtual power limiter system of the present invention is applied. FIG. 12 is a block diagram showing one example of a pneumatic cylinder system to which the virtual power limiter system of the present invention is applied.

A motor control system 40 shown in FIG. 11 includes: a motor M and a motor driver MD, which become the user's controlled object 3 ; a controller C which is open to the user and is used for inputting a command from the user's control strategy 5 to the motor M; and the virtual power limiter system 100 of the present invention. Here, a commercially available motor driver can be appropriately used as the motor driver MD.

The virtual power limiter system 100 includes a virtual power monitor system 101 and a virtual power limiter 102 and is constructed in the form of hardware made of an analog or digital electronic circuit, in the form of software including a program stored in an arithmetic unit of a microcomputer or the like, or in the form of combination of these. The system configuration and the other details of the virtual power limiter system 100 are the same as those described on the basis of FIG. 8 and FIG. 9 in the second embodiment i). To the user's control strategy 5 in the controller C are inputted an arbitrary input S from the user and an arbitrary feedback from the output of the motor M. Moreover, to the virtual power limiter system 100 are inputted the user torque command u _usr from the user's control strategy 5 , the input y _d to the conservative control strategy 104 , and the output information y of the motor M. A torque command u is inputted to the user's controlled object 3 from the virtual power limiter system 100 . In this example, the output information y of the motor M is the angular velocity of the motor and information obtained by an angular velocity sensor 108 using a publicly known encoder, tachometer or the like is fed back to the virtual power monitor system 101 and the virtual power limiter 102 . Moreover, in this example, the input y _d to the conservative control strategy 104 is a target angular velocity.

The virtual power limiter 102 includes the user's control system tuning strategy 120 and the weights 122 b to 124 b (W _v , W _usr , and W _csv ) tuned by the user's control system tuning strategy 120 . Moreover, the user's control system tuning strategy 120 is connected to the output P _v of the virtual power monitor 110 . To the weights 122 b to 124 b (W _v , W _usr , and W _csv ) are inputted the output y of an object to be tuned, the user torque command u _usr from the user's control strategy 5 , and the control input u _csv from the conservative control strategy 104 .

The configuration of a pneumatic cylinder system 80 shown in FIG. 12 is also the same as that of a motor control system 40 shown in FIG. 11. Here, the user's controlled object 3 is a pneumatic cylinder AC and respective commands are pressure commands. Positive or negative pressure proportional to a pressure command value (current) is applied to the pneumatic cylinder AC from a valve V 1 to extend or contract a piston P.

In this example, the output information y of the pneumatic cylinder AC is the volume change velocity ΔV of the cylinder and the information of the volume change velocity ΔV is appropriately fed back to the virtual power monitor system 101 and the virtual power limiter 102 by a sensor 107 . Moreover, in this example, the input y _d to the conservative control strategy 104 is a target volume change velocity.

Next, the operation of the motor control system 40 in accordance with this example will be described with reference to FIG. 11.

First, the user inputs a control input (arbitrary input S), for example, relating to the angular velocity of the motor with time on a horizontal axis from the controller C. Usually, if the user's control system 2 ′ is in a stable state, the user torque command u _usr by the user's control strategy 5 , which is outputted from the controller C, is applied as an input to the user's controlled object 3 (torque command) u=u _usr to the motor driver MD without being subjected to a new processing in the virtual power limiter system 100 , whereby the motor M is controlled and operated. At this time, the weight 123 b (W _usr ) of the virtual power limiter 102 is W _usr =1. In this manner, if the stability of the user's control system 2 ′ is held only by u _usr , the control input u _csv from the conservative control strategy 104 is not necessarily inputted to the motor M. Therefore, if the input u _usr by the user′ control strategy 5 does not need the control input u _usr from the conservative control strategy 104 and the user's control system 2 ′ is stable, the role of the conservative control strategy 104 can be limited to inputting the control input u _csv to the virtual power monitor 110 by setting the weight 124 b (W _csv ) to 0. Even in this case, the output information y of the motor. M and the control input u _csv from the conservative control strategy 104 are inputted to the virtual power monitor 110 and the virtual power monitor 110 keeps monitoring the stability of the user's control system 2 ′.

On the other hand, when the user's control system 2 ′ becomes unstable, if the virtual power limiter 102 is appropriately operated, the output y of the user's control system 2 ′ can be made asymptotically stable. In response to the output P _v of the virtual power monitor 110 , the virtual power limiter 102 operates the user's control system tuning strategy 120 connected to the virtual power monitor 110 and then determine the respective weights 122 b to 124 b (W _v , W _usr and W _csv ) to make the output y of the user's control system 2 ′ stable while keeping control.

In this manner, according to the motor control system 40 in accordance with the present invention, it is possible to always guarantee the asymptotical stability of the output y of the user's control system 2 ′.

SECOND EXAMPLE

Applications to Electric/Electronic Control System

FIG. 13 shows one example to which the present invention is applied to an electric/electronic control system. The electric/electronic control system is not necessarily limited to a system for controlling a rotary electric machine as shown in the first example, but can be assumed to be other various systems such as a power supply system for communications.

An electric/electronic control system 50 in accordance with this second example includes: an electric/electronic controlled object EE to become the user's controlled object 3 ; a controller C which is open to the user and is used for inputting a command from the user's control strategy 5 to the electric/electronic controlled object EE; and the virtual power limiter system 100 of the present invention. The electric/electronic controlled object EE includes a control voltage source VS, an ammeter AM, and a desired load R. The virtual power limiter system 100 includes the virtual power monitor 101 and the virtual power limiter 102 and, as is the case with the first example, is constructed of hardware, software, or a combination of them.

The basic configuration and operation of the electric/electronic control system 50 in accordance with the second example is the same as those of the motor control system 40 in the first example.

The control voltage source VS is controlled according to an input (voltage command) u to the user's controlled object 3 from the virtual power limiter system 100 to tune current passing through the desired load R. Current y passing through the desired load R is detected by the ammeter AM or a shunt resistance and is inputted to the gain 122 c (−K _v ) of the negative output feedback, the conservative control strategy 104 , and the virtual power monitor 110 in the virtual power limiter system 100 . At the same time, the virtual power monitor system 110 has also inputted the control input u _csv from the conservative control strategy 104 and monitors the stability of the user's control system 2 ′ by the use of y and u _csv .

To the user's control strategy S in the controller C are inputted an arbitrary input S from the user and an arbitrary feedback from the output of the electric/electronic controlled object EE. Moreover, to the virtual power limiter system 100 are inputted the user's voltage command u _usr from the user's control strategy 5 , the input y _d to the conservative control strategy 104 , and the output information y of the electric/electronic controlled object EE. Moreover, in this example, the input y _d to the conservative control strategy 104 is a target current. The target current value is, for example, a rated current value.

Next, the operation of the electric/electronic control system 50 in accordance with this second example will be described with reference to FIG. 13.

Usually, if the electric/electronic control system 50 is in a stable state, the user's voltage command u _usr from the controller C is inputted to the control voltage source VS as a voltage command u=u _usr to the user's controlled object 3 without being subjected to any new processing in the virtual power limiter system 100 , whereby the electric/electronic controlled object EE is controlled. That is, when a current output passing through the desired load R is not sufficient, the user's voltage command u _usr is increased by the feedback control between the user's control strategy 5 and the electric/electronic controlled object EE to increase the voltage command u to the electric/electronic controlled object EE to increase the current output, whereas when a current output is excessive, the user's voltage command u _usr is decreased to decrease the voltage command u to the electric/electronic controlled object EE to decrease the current output.

On the other hand, when the user's control system 2 ′ becomes unstable, if the virtual power limiter 102 is appropriately operated, the output y of the user's control system 2 ′ can be made asymptotically stable. In response to the output P _v of the virtual power monitor 110 , the virtual power limiter 102 operates the user's control system tuning strategy 120 connected to the virtual power monitor 110 and then determine the respective weights 122 b to 124 b (W _v , W _usr , and W _csv ) to make the output y of the user's control system 2 'stable while keeping control. Here, the user's control system tuning strategy 120 is a tuning strategy of the respective weights of the virtual power limiter 102 .

In this manner, according to the electric/electronic control system 50 in accordance with the present invention, it is possible to always guarantee the asymptotical stability of the output y of the user's control system 2 ′.

THIRD