Trajectory generation apparatus, multi-link system, and trajectory generation method

Patent metadata

Field	Value
Publication number	US-2021237270-A1
Application number	US-202117166817-A
Country	US
Kind code	A1
Filing date	Feb 3, 2021
Priority date	Feb 5, 2020
Publication date	Aug 5, 2021
Grant date	—

Abstract

In a trajectory generation apparatus, position coordinates of an obstacle existing in a motion space of a robot arm is acquired. A hand position at a second time, which is a time next to a first time, is estimated by using a learning result of machine learning, based on the position coordinates of the obstacle, a subject joint state of the robot arm at the first time, and a target joint state of the robot arm. A non-interfering joint state of the robot arm at which the obstacle does not interfere with the robot arm at the second time is searched for by using the hand position as a restriction.

First claim

What is claimed is:

1 . A trajectory generation apparatus configured to generate a trajectory of a multi-link mechanism in a motion space, the multi-link mechanism being a machine mechanism in which a plurality of links are connected by joints that are movable parts,
the trajectory generation apparatus comprising: an information acquisition unit configured to acquire obstacle information regarding a position of an obstacle existing in the motion space of the multi-link mechanism; an end position estimation unit configured to estimate a next end position that is an end position at a second time that is a time next to a first time, the end position being a position of an end effector of the multi-link mechanism,
based on (i) a subject joint state of the multi-link mechanism at the first time, (ii) a target joint state of the multi-link mechanism, and (iii) the obstacle information acquired by the information acquisition unit,
by using a learning result of a machine learning in which
a start joint state of the multi-link mechanism, a goal joint state of the multi-link mechanism, and the obstacle information are input and
the end position being the position of the end effector of the multi-link mechanism is output; and

a restriction search unit configured to search for a non-interfering joint state of the multi-link mechanism, in which the obstacle and the multi-link mechanism at the second time do not interfere with each other, while considering as a restriction the next end position estimated by the end position estimation unit.

2 . The trajectory generation apparatus according to claim 1 , further comprising:
a target arrival determination unit configured to determine whether the non-interfering joint state searched by the restriction search unit has reached the target joint state, wherein: until the non-interfering joint state searched by the restriction search unit is determined to have reached the target joint state by the target arrival determination unit, a series of processes of estimating the next end position by the end position estimation unit and searching for the non-interfering joint state by the restriction search unit are repeated while sequentially advancing the time.

3 . The trajectory generation apparatus according to claim 1 , wherein
the restriction search unit is configured to include: an end orientation generation unit configured to generate an end orientation that is an orientation of the end effector to take the next end position estimated by the end position estimation unit; a joint state obtainment unit configured to obtain a joint state of the multi-link mechanism satisfying the next end position estimated by the end position estimation unit and the end orientation generated by the end orientation generation unit; and an interference determination unit configured to determine whether an interference is present between the obstacle and the multi-link mechanism that takes the joint state obtained by the joint state obtainment unit, wherein: in response to the interference determination unit determining that the interference is not present, the restriction search unit is configured to end searching for the non-interfering joint state; and in response to the interference determination unit determining that the interference is present, the restriction search unit is configured to determine whether a counted number of times of end orientation generation, which is a counted number of times the end orientation is regenerated by the end orientation generation unit, is less than or equal to a specified number, wherein: in cases that it is determined that the counted number of times of end orientation generation is less than or equal to the specified number in response to the interference determination unit determining that the interference is present,
the end orientation is regenerated into a different end orientation by the restriction search unit with the end orientation generation unit, and searching for the non-interfering joint state is continued by the restriction search unit, and
in cases that it is determined that the counted number of times of end orientation generation is greater than the specified number in response to the interference determination unit determining that the interference is present,
searching for the non-interfering joint state while considering as the restriction the next end position is discontinued by the restriction search unit,
the next end position is re-estimated into a different next end position by the end position estimation unit, and
then searching for the non-interfering joint state while considering as the restriction the different next end position is re-started by the restriction search unit.

4 . The trajectory generation apparatus according to claim 3 , wherein:
the end orientation generation unit is configured to generate the end orientation at the second time for taking the next end position estimated by the end position estimation unit, such that a change of the end position at the second time from the orientation of the end effector at the first time is within a predetermined range.

5 . The trajectory generation apparatus according to claim 3 , further comprising:
a random number generation unit configured to add, as a noise, a random number to either (i) at least one of three based on which the next end position is estimated by the end position estimation unit, a first of the three being the obstacle information, a second of the three being the subject joint state, and a third of the three being the target joint state, or (ii) the next end position estimated by the end position estimation unit.

6 . The trajectory generation apparatus according to claim 3 , wherein:
the joint state obtainment unit is configured to obtain the joint state of the multi-link mechanism that satisfies the next end position estimated by the end position estimation unit and the end orientation generated by the end orientation generation unit, by using an inverse kinematics calculation.

7 . The trajectory generation apparatus according to claim 3 , wherein:
the joint state obtainment unit is configured to obtain the joint state of the multi-link mechanism satisfying the next end position estimated by the end position estimation unit and the end orientation generated by the end orientation generation unit, by using the learning result of machine learning in which
a combination of a position and an orientation that are enabled to be taken by the end effector is enabled to take is input and
the joint state of the multi-link mechanism is output.

8 . The trajectory generation apparatus according to claim 1 , further comprising:
a sub-target estimation unit configured to estimate a sub-target state that is a target joint state as a sub-goal until reaching a final target joint state of the multi-link mechanism, by
(i) using a learning result of machine learning in which the start joint state of the multi-link mechanism, the goal joint state of the multi-link mechanism, and the obstacle information of the obstacle are input, and a probability distribution of the joint state of the multi-link mechanism at an optional time, and
(ii) clustering with respect to the probability distribution of the joint state of the multi-link mechanism at the optional time, the probability distribution being obtained based on an initial joint state of the multi-link mechanism, the final target joint state of the multi-link mechanism, and the obstacle information acquired by the information acquisition unit,
wherein: a series of processing of estimating the next end position by the end position estimation unit and searching for the non-interfering joint state by the restriction searc

Assignees

• Denso Corp

Inventors

• Minoya Kenichi

Classifications

• B25J9/1664 Primary

  characterised by motion, path, trajectory planning · CPC title

• B25J9/163

  learning, adaptive, model based, rule based expert control · CPC title

• G05B2219/40371

  Control trajectory to avoid joint limit as well as obstacle collision · CPC title

• B25J9/1666 Primary

  Avoiding collision or forbidden zones · CPC title

• B25J13/089

  Determining the position of the robot with reference to its environment · CPC title