Human-robot collaboration
A human-robot collaboration system, including at least one processor; and a non-transitory computer-readable storage medium including instructions that, when executed by the at least one processor, cause the at least one processor to: predict a human atomic action based on a probability density function of possible human atomic actions for performing a predefined task; and plan a motion of the robot based on the predicted human atomic action.
1 . A human-robot collaboration system, comprising:
at least one processor; and
a non-transitory computer-readable storage medium including instructions that, when executed by the at least one processor, cause the at least one processor to:
predict a human atomic action by applying a probability density function over a set of possible human atomic actions associated with a predefined task, wherein each human atomic action is defined by pre-conditions specifying a required stable configuration of an element in a scene and post-conditions specifying a resulting stable configuration of the element in the scene after completion of that action;
plan a motion of the robot based on the predicted human atomic action; and
generate an instruction to cause the robot to perform the motion.
2 . The human-robot collaboration system of claim 1 , wherein the predefined task is defined by a state machine comprising:
a set of categorical states of stable configurations of elements in a scene, wherein the possible human atomic actions navigate between the categorical states; and
a dynamic probability density function of the possible human atomic actions at respective categorical states of the state machine.
3 . The human-robot collaboration system of claim 2 , wherein the instructions further cause at least one processor to:
parameterize the dynamic probability density function based on crowdsourced data of behavioral patterns exhibited by humans performing the predefined task.
4 . The human-robot collaboration system of claim 2 , wherein the instructions further cause the at least one processor to:
parameterize the dynamic probability density function based on a prior human atomic action.
5 . The human-robot collaboration system of claim 2 , wherein the dynamic probability density function is non-conservative.
6 . The human-robot collaboration system of claim 1 , wherein a sequence of human atomic actions of the predefined task depends on the predefined task's collaboration mode.
7 . The human-robot collaboration system of claim 2 , wherein the instructions further cause the at least one processor to:
generate the dynamic probability density function of the possible human atomic actions using a generative model.
8 . The human-robot collaboration system of claim 7 , wherein the generative model considers a prior human atomic action.
9 . The human-robot collaboration system of claim 7 , wherein the generative model comprises an obstacle avoidance factor.
10 . The human-robot collaboration system of claim 7 , wherein the robot is an autonomous vehicle and the generative model predicts non-motorized road user behavior.
11 . A non-transitory computer readable medium having stored thereon instructions that, when executed by at least one processor, cause the at least one processor of a human-robot collaboration system to:
predict a human atomic action by applying a probability density function over a set of possible human atomic actions associated with a predefined task, wherein each human atomic action is defined by pre-conditions specifying a required stable configuration of an element in a scene and post-conditions specifying a resulting stable configuration of the element in the scene after completion of that action;
plan a motion of the robot based on the predicted human atomic action; and
generate an instruction to cause the robot to perform the motion.
12 . The non-transitory computer readable medium of claim 11 , wherein the predefined task is defined by a state machine comprising:
a set of categorical states of stable configurations of elements in a scene, wherein the possible human atomic actions navigate between the categorical states; and
a dynamic probability density function of the possible human atomic actions at respective categorical states of the state machine.
13 . The non-transitory computer readable medium of claim 12 , wherein the instructions further cause the at least one processor to:
parameterize the dynamic probability density function based on crowdsourced data of behavioral patterns exhibited by humans performing the predefined task.
14 . The non-transitory computer readable medium of claim 12 , wherein the instructions further cause the at least one processor to:
parameterize the dynamic probability density function based on a prior human atomic action.
15 . The non-transitory computer readable medium of claim 12 , wherein the dynamic probability density function is non-conservative.
16 . The non-transitory computer readable medium of claim 11 , wherein a sequence of human atomic actions of the predefined task depends on the predefined task's collaboration mode.
17 . The non-transitory computer readable medium of claim 11 , wherein the instructions further cause the at least one processor to:
generate dynamic probability density function of the possible human atomic actions using a generative model.
18 . The non-transitory computer readable medium of claim 17 , wherein the generative model considers a prior human atomic action.
19 . The non-transitory computer readable medium of claim 17 , wherein the generative model comprises an obstacle avoidance factor.
20 . The non-transitory computer readable medium of claim 17 , wherein the robot is an autonomous vehicle and the generative model predicts non-motorized road user behavior.
21 . A human-robot collaboration system, comprising:
a prediction means for predicting a human atomic action by applying a probability density function over a set of possible human atomic actions associated with a predefined task, wherein each human atomic action is defined by pre-conditions specifying a required stable configuration of an element in a scene and post-conditions specifying a resulting stable configuration of the element in the scene after completion of that action;
a planning means for planning a motion of the robot based on the predicted human atomic action; and
instruction generation means for generating an instruction to cause the robot to perform the motion.
22 . The human-robot collaboration system of claim 21 , wherein the predefined task is defined by a state machine comprising:
a set of categorical states of stable configurations of elements in a scene, wherein the possible human atomic actions navigate between the categorical states; and
a dynamic probability density function of the possible human atomic actions at respective states of the state machine.
23 . The human-robot collaboration system of claim 22 , further comprising:
a parameterization means for parameterizing the dynamic probability density function based on crowdsourced data of behavioral patterns exhibited by humans performing the predefined task.
24 . The human-robot collaboration system of claim 1 , wherein the predefined task is represented using a quasi-probabilistic Petri network that maintains multiple concurrent state beliefs via probabilistic tokens.