Method and control device for avoiding collisions between cooperating robots

• US 6,678,582 B2
• Filed: 05/30/2002
• Issued: 01/13/2004
• Est. Priority Date: 05/30/2002
• Status: Expired due to Term

• Current Assignee: KUKA Laboratories GmbH (Midea Group Co., Ltd)
• Original Assignee: KUKA Roboter GmbH (Midea Group Co., Ltd)
• Inventors: Waled, El-Houssaine
• Primary Examiner(s): Cuchlinski, Jr., William A.
• Assistant Examiner(s): Marc, McDieunel, Marc, McDieunel
• Application Number:
  US10/159,701
• Publication Number:
  US 20030225479A1
• Time in Patent Office: 593 Days
• Field of Search: 700/245, 700/251, 700/262, 700/263, 700/260, 700/254, 700/259, 700/247, 700/258, 318/568.11, 318/568.24, 318/646, 318/648, 318/560, 318/632, 318/580, 318/587, 318/568.12, 701/50, 701/211, 701/207, 701/212, 701/200, 701/208, 701/23, 701/213, 701/217, 701/221, 701/301, 701/214, 701/215, 901/1, 901/9, 901/15, 901/14
• 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 such as another robot is provided in which the user does not need to make any provisions in a robot program for avoiding collisions and defining common work-areas. Furthermore, the 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 components of a multi-axial industrial robot and at least one other object, said method comprising repeatedly:

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

forecasting a configuration of a trajectory of the components of the multi-axial industrial robot at said stopping times;

checking the predicted configuration through distance/interference algorithms for interference of the multi-axial industrial robot components with components of the at least at least one other obiect; 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 the trajectory.

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

4. A method according to claim 1 or 2 or 3, wherein interference checking involves determining distances between geometric approximations of components of the robot and the at least at least one other object.

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 claim 1, wherein forecasting the trajectory of a joint involves application of a relationship:

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

${\stackrel{.}{\mathrm{\theta <br><br>}}}_i=c_f·<br><br>\underset{k=0}{\overset{4}{\sum <br><br>}}<br><br>\text{\hspace{1em}<br><br>}<br><br>c_k·<br><br>{\mathrm{\theta <br><br>}}_{i,-k},$

where θ

<sub>i,−

k</sub>=θ

_i(t<sub>−

k</sub>), t<sub>−

k</sub>=t−

k·

Δ

t, and

c_f=1/(12 Δ

t), c₀=25, c₁=−

48, c₂=36, c₃=−

16, c₄=3.

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

${\mathrm{\gamma <br><br>}}_i=c_f·<br><br>\underset{k=0}{\overset{4}{\sum <br><br>}}<br><br>\text{\hspace{1em}<br><br>}<br><br>c_k·<br><br>{\stackrel{.}{\mathrm{\theta <br><br>}}}_{i,-k}$

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

determining a stopping distance of a joint i from ${\mathrm{\theta <br><br>}}_{i,\mathrm{stop}}=\frac{{\stackrel{.}{\mathrm{\theta <br><br>}}}_{i,0}^2}{\left(2<br><br>\text{\hspace{1em}<br><br>}<br><br>{\mathrm{\gamma <br><br>}}_{i,\max }\right\}}$

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 the at least one other obiect at said configuration.

10. A method according to claim 1, wherein a filter is applied to a 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 of claim 1, 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 claim 1, 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 cooperative control 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.

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 claim 1, wherein prior to an allowed contact between the robot and an object, a link is established causing the robot 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.

23. A control device for avoiding collisions between components of a multi-axial industrial robot and at least one other object, the control device comprising:

means for repeatedly 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 of the multi-axial industrial robot;

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

means for repeatedly checking the predicted configuration through distance/interference algorithms for interference of the robot components with components of the at least at least one other object; and

means for stopping the multi-axial industrial robot and/or the at least one other object 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 in connection with the realtime operating system provides 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 run an interpretation program for communication between the PC and the real-time operating systems.

27. A control device according to claim 24 or 25 or 26, wherein said processor means run respective parts of a program implementing a collision avoidance method.

28. A control device according to claim 24 or 25 or 26, 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.