Motion planning and control for systems with multiple mobile objects

• US 6,004,016 A
• Filed: 08/06/1996
• Issued: 12/21/1999
• Est. Priority Date: 08/06/1996
• Status: Expired due to Fees

• Current Assignee: Northrop Grumman Systems Corporation (Northrop Grumman Corporation)
• Original Assignee: TRW Inc. (Northrop Grumman Corporation)
• Inventors: Spector, Victor A.
• Primary Examiner(s): Moise, Emmanuel L.
• Assistant Examiner(s): Chase, Shelly A
• Application Number:
  US08/692,733
• Time in Patent Office: 1,232 Days
• Field of Search: 364/167.01, 364/433, 364/474.28, 364/167.02, 701/301, 395/82, 395/90
• US Class Current: 700/56
• CPC Class Codes: B25J 9/1666 Avoiding collision or forbidden zones
• CPC Class Codes: G05B 2219/39083 Robot interference, between two robot arms
• CPC Class Codes: G05B 2219/40429 Stochastic, probabilistic generation of intermediate points
• CPC Class Codes: G05B 2219/40448 Preprocess nodes with arm configurations, c-space and planning by connecting nodes
• CPC Class Codes: G05B 2219/40474 Using potential fields

Abstract

A method and apparatus for path planning and execution of movements of multiple mobile objects, such as robotic manipulators (64, 66), in a common workspace. Path planning at a relatively coarse scale yields, for each object, path definitions (24) in configuration space (c-space), which is an n-dimensional space, where there are n degrees of freedom of movement of each object. The coarse path definitions are interpolated to provide primary control signals (54) to execute object movements, in combination with collision avoidance control signals (56) derived from an artificial force field model (46) that generates repulsion forces based on mutual proximity of the objects. Path planning includes determining which subregions or cells of c-space are subject to potential collision, and selecting multiple trial path segments until a path is found around those cells. For execution, the coarse scale path parameters, are expanded to a finer scale by interpolation (40), and the artificial force field model generates additional control signals (56) corresponding to the repulsion forces needed to maintain separation of the objects in the workspace.

Claims

1. A system for controlling the movements of multiple mobile objects through a workspace, wherein the position of each object is dependent on the configuration of a plurality of interconnected linkages, as defined by a plurality, n, of configuration settings corresponding to n degrees of freedom of the object, the system comprising:

a path planner for planning in advance a series of movements of each of the mobile objects to provide a relatively coarse scale plan for the movements, wherein the path planner determines paths in a configuration space defined by n orthogonal axes corresponding to n configuration settings that define the position of each object in three-dimensional space; and

a path execution module operable to move the objects in accordance with the relatively coarse scale plan provided by the path planner, and including a fine-scale artificial force field collision avoidance subsystem, to provide control signals to the movable objects such that they are moved along paths determined in part by the path planner and in part by the collision avoidance subsystem, whereby the objects are moved from starting points to desired endpoints without collisions with each other or with any stationary obstacles, and wherein the path execution module includes

a waypoint interpolator, to generate finer resolution paths from path data provided by the path planner.

a control law module for generating control signals based on the finer resolution path through configuration space and predicted configuration setting of the object,

a collision avoidance force model, for generating collision avoidance control signals based on mutual proximity of the objects in the workplace, and

means for combining the collision avoidance control signals and the control signals from the control law module, wherein collision avoidance is achieved on a fine scale by modifying the paths planed through the configuration space.

2. A system as defined in claim 1, wherein the path planner includes:

a configuration-space database defining the workspace as multiple n-dimensional cells;

database generation logic, for determining a collision status for each cell of the configuration-space database, wherein the collision status is either "full," indicating that there is a potential for collision for all combinations of configuration settings within the cell, or "free," indicating that there is no potential for collision in the cell, or "mixed," indicating that there is potential for collision for some combinations of configuration settings within the cell; and

a path planning executive module using information about each cell of the configuration-space database, to plan a path from one point in configuration space to another, in such a manner that collisions are avoided.

3. A system as defined in claim 2, wherein the path planner further includes:

a path segment timing module, for generating time data for the path planning executive module, based on given dynamic constraints to be applied to joint variables.

4. A system as defined in claim 3,

wherein the waypoint interpolator includes a path segment timing module, for generating time data to be used in planning the interpolated path, based on given dynamic constraints to be applied to joint variables.

5. A system as defined in claim 2, wherein the path planning executive module further includes:

means for selecting a first trial path and determining whether the path passes through any full or mixed cells; and

means for systematically selecting a second and other trial paths, if the earlier selected trial paths are subject to collision, wherein a collision-free path is ultimately selected.

6. A system as defined in claim 5, wherein the means for systematically selecting a second and other trial paths includes:

means for detecting a midpoint in a portion of the trial path that passes through a collision region of full or mixed cells; and

means for selecting a trial waypoint located on a line through the midpoint and orthogonal to the trial path that passes through the midpoint, wherein the trial waypoint is selected to be on the planned path if the trial waypoint is not in the collision region.

7. A system as defined in claim 1, wherein the path execution module includes:

an object dynamic model, for providing estimates of object configuration settings for use by the control law module and the collision avoidance force model.

8. A system for controlling the movements of multiple robotic manipulators through a workspace, wherein the position of each manipulator is dependent on the configuration of a plurality of interconnected linkages, as defined by a plurality, n, of configuration joint settings corresponding to n degrees of freedom of the manipulator, the system comprising:

a path planner for planning in advance a series of movements of each of a plurality robotic manipulators to provide a relatively coarse scale plan for the movements, wherein the path planner determines paths in a configuration space defined by n orthogonal axes corresponding to n configuration joint settings that define the position of each manipulator in three-dimensional space; and

a path execution module operable to move the manipulators in accordance with the relatively coarse scale plan provided by the path planner, and including a fine-scale artificial force field collision avoidance subsystem, to provide control signals to the manipulators such that they are moved along paths determined in part by the path planner and in part by the collision avoidance subsystem, whereby the manipulators are moved from starting points to desired endpoints without collisions with each other or with any stationary obstacles, and wherein the path execution module includes

a waypoint interpolator, to generate finer resolution paths from path data provided by the path planner,

a control law module for generating control signals based on the finer resolution paths through configuration space and predicted states of the manipulators;

a collision avoidance force model, for generating collision avoidance control signals based on mutual proximity of the manipulators and any other objects in the workspace, and

means for combining the collision avoidance control signals and the control signals from the control law module, wherein collision avoidance is achieved by modifying the paths planned through the configuration space.

9. A system as defined in claim 8, wherein the path planner includes:

a configuration-database defining the workplace as multiple n-dimensional cells;

database generation logic, for determining a collision status for each cell of the configuration-space database, wherein the collision status is either "full," indicating that there is a potential for collision for all combinations of configuration joint settings within the cell, or "free," indicating that there is no potential for collision in the cell, or "mixed," indicating that there is potential for collision for some combinations of configuration joint settings within the cell; and

a path planning executive module using information about each cell of the configuration-space database, to plan a path from one point in configuration space to another, in such a manner that collisions are avoided.

10. A system as defined in claim 9, wherein the path planner further includes:

a path segment timing module, for generating time data for the path planning executive module, based on given dynamic constraints to be applied to joint variables.

11. A system as defined in claim 10,

wherein the waypoint interpolator includes a path segment timing module, for generating time data to be used in planning the interpolated path, based on given dynamic constraints to be applied to joint variables.

12. A system as defined in claim 9, wherein the path planning executive module further includes:

means for selecting a first trial path and determining whether the path passes through any full or mixed cells; and

means for systematically selecting a second and other trial paths, if the earlier selected trial paths are subject to collision, wherein the a collision-free path is ultimately selected.

13. A system as defined in claim 12, wherein the means for systematically selecting a second and other trial paths includes:

means for detecting a midpoint in a portion of the trial path that passes through a collision region of full or mixed cells; and

means for selecting a trial waypoint located on a line through the midpoint and orthogonal to the trial path that passes through the midpoint, wherein the trial waypoint is selected to be on the planned path if the trial waypoint is not in the collision region.

14. A system as defined in claim 8, wherein the path execution module further includes:

a manipulator dynamic model, for providing estimates of manipulator configuration joint settings for use by the control law module and the collision avoidance force model.

15. A method for planning and executing movements of multiple mobile objects through a workspace, wherein the position of each object is dependent on the configuration of a plurality of interconnected linkages, as defined by a plurality, n, of configuration settings corresponding to n degrees of freedom of the object, the method comprising the steps of:

planning in advance a path having series of movements of each of the mobile objects to provide a relatively coarse scale plan for the movements, wherein the planning step determines paths in a configuration space defined by n orthogonal axes corresponding to n configuration settings that define the position of each object in three-dimensional space;

controlling the path execution in such a way as to move the objects primarily in accordance with the relatively coarse scale plan provided by the planning step, and including an additional step of generating collision avoidance object control signals based on an artificial force field model, wherein the step of controlling the path execution movement includes

interpolating between coarse path planning waypoints to generate finer resolution planned paths,

generating control signals based on the finer resolutions paths through configuration space and predicted configuration settings of the objects,

generating collision avoidance control signals based on mutual proximity of objects in the workspace, and

combining collision avoidance control signals and the control signals based on the finer resolution planned path, wherein collision avoidance is achieved by modifying the paths planned through the configuration space; and

applying control signals to the movable objects such that they are moved along paths determined in part by the path planning step and in part by the step of controlling the path execution, whereby the objects are moved from starting points to desired endpoints without collision with each other or with any stationary obstacles.

16. A method as defined in claim 15, wherein the path planning step includes:

setting up a configuration-space database defining the workspace as multiple n-dimensional cells;

determining a collision status for each cell of the configuration space database, wherein the collision status is either "full," indicating that there is a potential for collision for all combinations of configuration settings within the cell, or "free," indicating that there is no potential for collision in the cell, or "mixed," indicating that there is potential for collision for some combinations of configuration settings within the cell; and

generating a coarse path plan using information about each cell of the configuration-space database, to provide for movement in a path from one point in configuration space to another in such a manner that collisions are avoided.

17. A method as defined in claim 16, wherein the step of generating a coarse path includes:

generating time data for multiple segments of the coarse path, based on given dynamic contraints to be applied to joint variables.

18. A method as defined in claim 17,

wherein the interpolating step includes generating time data to be used in planning the interpolating path, based on dynamic constraints to be applied to joint variables.

19. A method as defined in claim 18, wherein the step of generating a coarse path plan includes:

selecting a first trial path and determining whether the path passes through any full or mixed cells; and

systematically selecting a second and other trial paths, if the earlier selected trial paths are subject to collision, wherein a collision-free path is ultimately selected.

20. A method as defined in claim 21, wherein the step of systematically selecting a second and other trial paths includes:

detecting a midpoint in a portion of the trial path that passes through a collision region of full or mixed cells; and

selecting a trial waypoint located on a line through the midpoint and orthogonal to the trial path that passes through the midpoint, wherein the trial waypoint is selected as being on the planned path if the trial waypoint is not in the collision region.

21. A method as defined in claim 15, wherein the steps of controlling the path execution further includes;

dynamic modeling movement of the objects, to provide estimates of object configuration settings for use in the step of generating control signals based on the finer resolution paths and in the step of generating collision avoidance control signals.

22. A method for planning and executing movements of multiple robotic manipulators through a workspace, wherein the position of each manipulator is dependent on the configuration of a plurality of interconnected linkages, as defined by a plurality, n, of configuration joint settings corresponding to n degrees freedom of the manipulator, the method comprising the steps of:

planning in advance a path having series of movements of each of the manipulators to provide a relatively coarse scale plan for the movements, wherein the planning step determines paths in a configuration space defined by n orthogonal axes corresponding to n configuration settings that define the position of each manipulator in three-dimensional space;

controlling the path execution in such a way as to move the manipulators primarily in accordance with the relatively coarse scale plan provided by the planning step, and including an additional step of generating collision avoidance object control signals based on an artificial force field model, wherein the step of controlling the path execution movement includes

interpolating between coarse path planning waypoints to generate finer resolution planned paths,

generating control signals based on the finer resolution paths through configuration space and predicted configuration settings of the manipulators,

generating collision avoidance control signals based on mutual proximity of manipulators and any obstacles in the workspace, and

combining the collision avoidance control signals and the control signals based on the finer resolution planned paths, wherein collision avoidance is achieved by modifying the paths planned through the configuration space; and

applying control signals to the manipulators in such that they are moved along paths determined in part by the path planning step and in part by the step of controlling the path execution, whereby the manipulators are moved from starting points to desired endpoints without collisions with each other or with any stationary obstacles.

23. A method as defined in claim 22, wherein the path planning step includes:

setting up a configuration-space database defining the workspace as multiple n-dimensional cell;

determining a collision status for each cell of the configuration space database, wherein the collision status is either "full," indicating that there is a potential for collision for all combinations of configuration settings within the cell, or "free," indicating that there is no potential for collision in the cell, or "mixed,"indicating that there is potential for collision for some combinations of configuration settings within the cell; and

generating a coarse path plan using information about each cell of the configuration-space database, to provide for movement in a path from one point in configuration space to another in such a manner that collisions are avoided.

24. A method as defined in claim 23, wherein the step of generating a course path includes:

generating time data for multiple segments of the coarse path, based on given dynamic constraints to be applied to joint variables.

25. A method as defined in claim 24,

wherein the interpolating step includes generating time data to be used in planning the interpolated path based on dynamic contraints to be applied to joint variables.

26. A method as defined in claim 23, wherein the step of generating a coarse path plan includes:

selecting a first trial path and determining whether the path passes through any full or mixed cells; and

systematically selecting a second and other trial paths, if the earlier selected trial paths are subject to collision, wherein a collision-free path is ultimately selected.

27. A method as defined in claim 26, wherein the step of systematically selecting a second and other trial paths include:

detecting a midpoint in a portion of the trial path that passes through a collision region of full or mixed cells; and

selecting a trial waypoint located on a line through the midpoint and orthogonal to the trial path passes through the midpoint, wherein the trial waypoint is selected as being on the planned path if the trial waypoint is not in the collision region.

28. A method as defined in claim 22, wherein the step of controlling the path execution further includes:

dynamically modeling movement of the manipulators, to provide estimates of manipulator configuration joint setting for use in the step of generating control signals based on the finer resolution paths and in the step of generating collision avoidance control signals.