METHOD AND CONTROL DEVICE FOR AVOIDING COLLISIONS BETWEEN COOPERATING ROBOTS

• US 20030225479A1
• Filed: 05/30/2002
• Published: 12/04/2003
• Est. Priority Date: 05/30/2002
• Status: Active Grant

• Current Assignee: KUKA Laboratories GmbH (Midea Group Co., Ltd)
• Original Assignee: KUKA Roboter GmbH (Midea Group Co., Ltd)
• Inventors: Waled, El-Houssaine
• Granted Patent: US 6,678,582 B2
• Time in Patent Office: Days
• US Class Current: 700/245
• CPC Class Codes: B25J 9/161 Hardware, e.g. neural networks, fuzzy logic, interfaces, processor
• CPC Class Codes: B25J 9/1676 Avoiding collision or forbidden zones
• CPC Class Codes: G05B 2219/34258 Real time system, qnx, works together with non real time system, windows nt
• CPC Class Codes: G05B 2219/39082 Collision, real time collision avoidance
• CPC Class Codes: G05B 2219/39097 Estimate own stop, brake time, then verify if in safe distance
• CPC Class Codes: G05B 2219/39271 Ann artificial neural network, ffw-nn, feedforward neural network

Abstract

A method of avoiding collisions between a robot and at least one other object, e.g. another robot is disclosed in which the user does not need to make any provisions in a robot program for avoiding collisions, e.g. defining common work-areas. Furthermore, the inventive method allows for automatic configuration of the workcell from a collision avoidance standpoint. It determines automatically which components have potential collisions with which other components. Since the inventive method is based on predicting the configurations of the moving components over a period of time sufficient enough to allow the machines to stop safely and checks for interference, a priori knowledge of trajectories is not required. If a collision is predicted the machines are commanded to a stop on or off their paths. In this way the inventive collision avoidance method can also be used as a safeguard with other explicit methods.

Claims

1. A method of avoiding collisions between a robot and at least one other object, said method comprising regularly:

determining a stopping time for an automatically or manually controlled robot movement on the basis of actual and past joint positions and velocities of each robot joint;

forecasting a configuration of a trajectory of the robot at said stopping times;

checking the predicted configuration through distance/interference algorithms for interference of robot components with components of said other objects; and

stopping the robot and/or said other objects in case a collision is imminent.

2. A method according to claim 1, wherein the robot is stopped on its path.

3. A method according to claim 1, wherein the robot is stopped off its path.

4. A method according to one of claims 1 to 3, wherein interference checking involves determining distances between geometric approximations of components of the robot and said other objects.

5. A method according to claim 4, wherein said geometric approximations have different levels of accuracy which are checked according to a desired accuracy in distance checking.

6. A method according to one of the preceding claims, wherein forecasting the trajectory of a joint involves application of a relationship:

θ

_i=½

γ

_it_i,s²+{dot over (θ

)}_i,0t_i,s+θ

_i,0, 

 

(1)

where θ

_i is a predicted position of joint i, where θ

_i,0 and {dot over (θ

)}_i,0 are a position and a velocity of joint i at a current time t, respectively, where t_i,s is the stopping time of joint i, and where γ

_i is an estimated current joint acceleration.

7. A method according to claim 6, wherein a velocity of joint i is estimated from the joint position at a current time and at earlier times using:

8. A method according to claim 7, wherein the acceleration of joint i is estimated from the joint velocity at a current time and at earlier times, using:

9. A method according to claim 8, wherein configuration forecasting and interference checking involves:

determining a stopping distance of joint i from

θ

_i,stop={dot over (θ

)}_i,0²/(2γ

_i,max),

where γ

_i,max is a maximum negative acceleration of the joint;

adding the stopping distance to the actual position of the joint to obtain a configuration over the stopping time; and

determining a distance between robot components and said other objects at said configuration.

10. A method according to one of the preceding claims, wherein a filter is applied to the joint position and velocity data used for position prediction.

11. A method according to claim 10, wherein said filter is applied in the time domain.

12. A method according to claim 11, wherein said filter is a running average.

13. A method according to one of the preceding claims, wherein configuration forecasting and interference checking involves using at least one neural network to estimate stopping times of the robot axes.

14. A method according to claim 13, wherein one neural network is used for each robot joint.

15. A method according to claim 14, wherein input to said neural networks includes one or more element(s) from a group comprising joint velocities, joint angles, joint positions, robot base positions, robot base velocities and tool or load masses.

16. A method according to one of the preceding claims, wherein all acquired information concerning collisions is stored in a workcell description file for communication to a real-time operating system responsible for controlling the robot movement.

17. A method according to claim 16, wherein the workcell description file contains pointers to geometric approximation files needed to describe all objects present in the workcell.

18. A method according to claim 17, wherein a copy of the workcell description file is maintained and updated on each controller, i.e. on each cooperative mechanism present in the workcell.

19. A method according to claim 18, wherein the workcell description file is updated at least when a geometric change happens in the workcell, e.g. mounting of a new tool to the robot.

20. A method according to claim 19, wherein the workcell description file including a description of the geometric change is updated in the real-time operating system.

21. A method according to one of the preceding claims, wherein prior to an allowed contact between the robot and an object, e.g. during a handling operation a link is established causing the robot and, where applicable, the object not to stop during a subsequent collision.

22. A method according to claim 21, wherein the link is broken after the allowed contact operation is terminated.

27. A control device according to one of claims 24 to 26, wherein said processor means are devised for running respective parts of a program implementing a collision avoidance method, in particular a method as described in preceding claims 1 to 22.

23. A control device for avoiding collisions between a robot and at least one other object, said control device comprising:

means for regularly determining a stopping time for an automatically or manually controlled robot movement on the basis of actual and past joint positions and velocities of each robot joint;

means for regularly forecasting a configuration of a trajectory of the robot at said stopping times;

means for regularly checking the predicted configuration through distance/interference algorithms for interference of robot components with components of said other objects; and

means for stopping the robot and/or said other objects in case a collision is imminent.

24. A control device according to claim 23, comprising processor means for running two operating systems in parallel, one of which being a standard PC operating system, the other one being a real-time operating system for controlling a robot motion.

25. A control device according to claim. 24, wherein said processor means at least in connection with the realtime operating system are designed for communication with real-time operating systems of other corresponding control devices present in the same workcell.

26. A control device according to claim 25, wherein said processor means are devised for running an interpretation program for communication between the PC and the real-time operating systems.

28. A control device according to one of claims 24 to 27, comprising memory units in conjunction with both operating systems for storing workcell related geometry data.

29. A control device according to claim 28, further comprising an input device for inputting workcell related geometric data.

30. A control device according to claim 29, wherein the input device is a disk drive or a network connection device.