Method for determining trajectory of multi-motor control system avoiding obstacle

Patent metadata

Field	Value
Publication number	US-9076337-B2
Application number	US-201314031175-A
Country	US
Kind code	B2
Filing date	Sep 19, 2013
Priority date	Sep 19, 2013
Publication date	Jul 7, 2015
Grant date	Jul 7, 2015

Abstract

The optimization of cost function representing a movement of the mass by the multi-motor control system from an initial point to a final point is subject to non-convex constraints due to avoidance of a region of the obstacle located between the initial point and the final point. Hence, the method determines a union of convex regions connecting the initial point with the final point, such that each convex region does not intersect the region of the obstacle, a convex constraint for each convex region and a set of switching times when the trajectory crosses boundaries of the convex regions. The method optimizes the cost function by jointly updating the positions and the set of switching times to produce an optimal trajectory of the movement. The optimizing is subject to the set of convex constraints applied independently for each corresponding convex region.

First claim

We claim:

1. A method for determining a trajectory for a multi-motor control system moving a mass, wherein the trajectory includes positions of the mass as a function of time, comprising:
determining a cost function representing a movement of the mass by the multi-motor control system from an initial point to a final point, wherein an optimization of the cost function is subject to non-convex constraints due to avoidance of a region of the obstacle located between the initial point and the final point;
determining a union of convex regions connecting the initial point with the final point, each convex region does not intersect the region of the obstacle;
determining a convex constraint for each convex region to produce a set of convex constraints;
determining a set of switching times when the trajectory crosses boundaries of the convex regions; and

optimizing the cost function by jointly updating the positions and the set of switching times to produce an optimal trajectory of the movement, wherein the optimizing is subject to the set of convex constraints, such that each convex constraint is applied independently for each corresponding convex region, wherein steps of the method are performed by a processor; selecting, for each convex region in the union, the convex constraint of the convex region from the set of convex constraints based on a time of the movement and the set of switching times;
optimizing, for each convex region in the union, the cost function subject to the convex constraint from the time of the movement defined by the switching times of the convex region;
updating the set of the switching times; and
repeating the selecting, the optimizing and the updating until a terminal condition is met.

2. The method of claim 1 , wherein the union of convex regions includes a first convex region and a second convex region, and a switching time corresponds to a time when the trajectory crosses a boundary between the first and the second convex regions, wherein a convex constraint of the first convex region is applied during a period of time the trajectory is within the first convex region, and a convex constraint of the second convex region is applied during a period of time the trajectory is within the second convex region.

3. The method of claim 1 , wherein the optimizing is performed iteratively until a gradient of the cost function with respect to the set of switch times is below a threshold.

4. The method of claim 1 , further comprising:
determining the convex constraints based on a feasible trajectory avoiding the obstacle;
initializing the set of switching times based on the feasible trajectory;

optimizing the cost function based on the initialized set of the switching times;
updating the set of switching times based on a set of gradients of the cost function with respect to the set of switching times; and
repeating the optimizing and the updating until each gradient is below a threshold.

5. The method of claim 4 , further comprising:
approximating the set of gradients based on a sensitivity equation.

6. The method of claim 5 , wherein the approximating comprises:
determining sensitivity equations according to the cost function, constraints, and time intervals specified by switching times, and
solving an initial value problem of the sensitivity equations.

7. The method of claim 4 , where the updating the set of switching times is bounded by an upper bound and a lower bound of the set of switching times estimated from a set of feasible trajectories avoiding the obstacle.

8. The method of claim 7 , where the updating set of switching times includes selecting the switching times from a finite set of values satisfying the upper and lower bounds of switch times.

9. The method of claim 7 , further comprising:
determining the set of feasible trajectories by optimizing a shortest moving time cost function or a quadratic function of decision variables.

10. The method of claim 1 , wherein the cost function includes an energy consumption of the multi-motor control system moving the mass according the trajectory.

11. The method of claim 1 , wherein the movement of the mass is subject to acceleration and velocity constraints of the multi-motor control system, such that the cost function is optimized subject to the acceleration and the velocity constraints.

12. A motion controller for controlling a multi-motor control system moving a mass according to a trajectory defining positions of the mass as a function of time from an initial point to a final point, such that the mass avoids at least one obstacle, wherein the multi-motor control system includes at least two motors for moving the mass into at least two directions, the controller comprising:
a trajectory generator module for determining a union of convex regions connecting the initial point with the final point, each convex region does not intersect a region of the obstacle, and for optimizing a cost function representing a movement of the mass by the multi-motor control system subject to a set of convex constraints to determine the trajectory, wherein each convex constraint corresponds to a convex region of the union, and wherein the optimizing applies each convex constraint independently for each corresponding convex region; and
a control module for generating a control signal to control the motors of the multi-motor control system according to the trajectory; determining a convex constraint for each convex region to produce the set of convex constraint and a set of switching times for crossing boundaries of the convex regions; and for optimizing jointly the cost function and the set of switching times, wherein the optimizing jointly comprises:
selecting, for each convex region in the union, the convex constraint of the convex region from the set of convex constraints based on a time of the movement and the set of switching times;
optimizing, for each convex region in the union, the cost function subject to the convex constraint for the time of the movement defined by the convex region;
updating the set of the switching limes; and
repeating the selecting, the optimizing and the updating until a gradient of the cost function with respect to the set: of switch times is below a threshold.

13. The motion controller of claim 12 , wherein the processor is configured for
determining the convex constraints based on a feasible trajectory avoiding the obstacle;
initializing the set of switching times based on the feasible trajectory;
optimizing the cost function based on the initialized set of the switching times to update the trajectory;
updating the set of switching times based on a set of gradients of the cost function with respect to the set of switching times; and
repeating the optimizing and the updating until each gradient is below a threshold.

14. The motion controller of claim 13 , wherein the set of switching times is bounded by an upper bound and a lower bound, and the processor selects the switching times from a finite set of values satisfying the upper and lower bounds of switch times.

15. The motion controller of claim 12 , wherein the cost function includes an energy consumption of the motors moving the mass according the trajectory.

16. A multi-motor control system for moving a mass according a trajectory, comprising:
at least two motors for moving the mass into at least two directions;
a motion controller for determining the trajectory defining positions of the mass as a function of time from an initial point to a final point, such that the mass avoids at least one obstacle, the motion controller com

Assignees

• Mitsubishi Electric Res Lab

Inventors

• Wang Yebin
• Yu Han
• Bortoff Scott A

Classifications

• G05B2219/40519

  Motion, trajectory planning · CPC title

• B25J9/1676 Primary

  Avoiding collision or forbidden zones · CPC title

• G01C21/00

  Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 (measuring distance traversed on the ground by a vehicle G01C22/00; control of position, course, altitude or attitude of vehicles G05D1/00; traffic control systems for road vehicles involving transmission of navigation instructions to the vehicle G08G1/0968) · CPC title

• G08G1/163

  involving continuous checking · CPC title

• G08G1/16 Primary

  Anti-collision systems (road vehicle drive control systems for predicting or avoiding probable or impending collision otherwise than by control of a particular sub-unit B60W30/08) · CPC title