Robot, robot action decision-making system and program thereof

The robot system optimizes task execution by estimating human behavior and selecting collaborative actions based on work costs, improving efficiency in human-robot collaboration.

JP2024162293APending Publication Date: 2024-11-21WASEDA UNIV
View PDF 2 Cites 0 Cited by

Patent Information

Application Number
JP2023077663
Authority / Receiving Office
JP · JP
Patent Type
Applications
Current Assignee / Owner
Filing Date
2023-05-10
Publication Date
2024-11-21

AI Technical Summary

Technical Problem

Existing autonomous mobile robots struggle to effectively collaborate with humans by selecting appropriate collaborative behaviors such as requesting or supporting actions based on the human's situation and intentions, leading to inefficiencies in task execution.

Method used

A robot system that includes a detection device to gather environmental information, a control device to estimate human behavior, and a motion planning unit to select actions based on collaborative work costs, allowing the robot to adapt its behavior to optimize task execution with humans.

Benefits of technology

The system enables the robot to estimate human behavior and select appropriate collaborative actions, enhancing the efficiency of task execution by aligning with human intentions and reducing overall work effort.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 2024162293000001_ABST
    Figure 2024162293000001_ABST
Patent Text Reader

Abstract

To allow action adaptive to the state of human movement when working with humans to carry out a series of works.SOLUTION: A robot 10 is equipped with a control device 13 for controlling action of the robot 10 so as to execute a series of works consisting of a plurality of tasks in cooperation with a co-worker. The control device 13 includes a co-worker movement estimation unit 24 that estimates movement of the co-worker with respect to the task based on changes over time in position information and speed information of the co-worker, and an action plan setting unit 25 that sets an action plan for the robot 10 according to the estimated movement of the co-worker. The action plan setting unit 25 creates action plans for each of action options consisting of normal action, request action, and support action, compares each action in each of the action options by obtaining a collaborative work cost corresponding to the load of the tasks performed by the robot and the co-worker, and thereby selects an appropriate action plan.SELECTED DRAWING: Figure 1
Need to check novelty before this filing date? Find Prior Art

Description

[Technical field]

[0001] The present invention relates to a robot that can perform actions adapted to the behavioral situation of a human when performing a series of tasks in collaboration with the human, a robot behavior decision system, and a program therefor. [Background technology]

[0002] Recently, various autonomous mobile robots capable of autonomously moving in complex environments where humans exist have appeared, and the demand for autonomous mobile robots that execute predetermined tasks in cooperation with humans is increasing. Such autonomous mobile robots need to sense the status of surrounding humans and obstacles and execute tasks while avoiding interference with these humans and obstacles. Furthermore, robots that move autonomously in environments where humans coexist are required to operate in cooperation with humans, taking into account the movements of the humans. Therefore, as the results of their own research, the present inventors have proposed various robots that perform action plans while searching for optimal movement routes and destinations that take into account the movements of humans in the environment in which the robot moves (see Patent Document 1, Patent Document 2, etc.). [Prior art documents] [Patent documents]

[0003] [Patent Document 1] JP 2020-46773 A [Patent Document 2] JP 2023-32381 A Summary of the Invention [Problem to be solved by the invention]

[0004] In a workplace where multiple people (workers) work together to perform a series of tasks, each worker usually performs the tasks assigned to him / her. However, sometimes, by taking collaborative concessionary actions such as "asking someone to do it" or "doing it for someone," it is possible to improve work efficiency. Such collaborative concessionary actions include requesting actions, which try to improve one's own efficiency at the expense of the other person's efficiency, and supporting actions, which try to improve the other person's efficiency at the expense of one's own efficiency. In addition, a proactive relationship that assertively encourages concessions with the other person and a reactive relationship that synchronizes with the other person's proactive actions can be considered. Therefore, in order to effectively improve the efficiency of the robot and the collaborator, even when a robot that can act freely in a complex environment where humans coexist collaborates with humans, it is necessary for the robot to select requesting actions or supporting actions according to each other's situation and the other person's behavioral intention, and to sequentially judge the proactive and reactive relationships, just as in the case of human-to-human tasks.

[0005] The present invention has been devised with an eye on these problems, and its purpose is to provide a robot, a robot behavior decision-making system, and a program for the same, which predict the behavior of a collaborator when working collaboratively with the robot, plans actions including requests and assistance to the other party in addition to normal actions, and selects from these an appropriate action depending on the situation. [Means for solving the problem]

[0006] In order to achieve the above object, the present invention provides a robot which mainly comprises an operating unit which operates to be able to execute a predetermined task, a detection device which detects surrounding environmental information, and a control device which controls the operation of the operating unit based on the detection result of the detection device, and which executes a series of operations consisting of a plurality of the tasks in cooperation with a collaborator, the detection device detects position information and speed information of objects around the robot including the collaborator, the control device includes a collaborator behavior estimation unit which estimates the behavior of the collaborator with respect to the task based on changes over time in the position information and the speed information of the collaborator, and a motion plan setting unit which sets a motion plan including a movement trajectory of the robot in accordance with the behavior of the collaborator estimated by the collaborator behavior estimation unit. and an action command unit that controls the action unit so that the robot operates in accordance with the action plan. The action plan setting unit creates the action plans for each of action options consisting of a normal action in which the robot and collaborator execute the task given to each other in advance, a request action in which the robot asks the collaborator to take over at least a part of the task of the robot, and a support action in which the robot takes over at least a part of the task of the collaborator, calculates a collaborative work cost corresponding to the load of the task to be performed by the robot and collaborator, compares each action in the action options based on the collaborative work cost, and selects the action plan for an appropriate action. Effect of the Invention

[0007] According to the present invention, the behavior of a collaborator performing a preset task can be estimated at any time, and based on the estimation, an appropriate robot behavior is selected from among those behaviors, including collaborative and yielding actions such as request actions and support actions in addition to normal actions, according to the collaborator and the work situation, enabling smooth cooperative work between the robot and collaborator. [Brief description of the drawings]

[0008] [Figure 1] FIG. 2 is a block diagram that shows a schematic configuration related to the motion control of the robot according to the present embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0009] Hereinafter, an embodiment of the present invention will be described with reference to the drawings.

[0010] FIG. 1 is a block diagram showing only the configuration related to the operation control of the robot according to this embodiment. The robot 10 according to this embodiment is an autonomous mobile robot that operates to be able to execute a series of tasks set in advance while cooperating with surrounding humans (hereinafter referred to as "collaborators"). The series of tasks includes various tasks that are actions of moving to a predetermined target and executing the predetermined tasks. For example, when the robot 10 and the collaborator share a series of tasks of collecting objects such as indoor garbage scattered in multiple locations to a trash can or the like that is a destination point in another location, the robot 10 and the collaborator each have a task of moving toward the object at a target location that is a different location, a task of each of them gripping each object, and a task of transporting the object to another target location that is a destination point for each of them.

[0011] The robot 10 is provided with an operating unit 11 consisting of mechanisms and devices configured to enable various operations including the execution of preset tasks, a detection device 12 that detects environmental information around the robot 10, and a control device 13 that controls the operation of the operating unit 11 based on the detection results of the detection device 12.

[0012] The operating unit 11 includes a moving mechanism 15 including a cart or the like for moving the robot 10 using a driving device such as a motor as a power source, a speaker 16 that emits a predetermined sound, and an arm 17 that can move within a predetermined space. The moving mechanism 15, speaker 16, and arm 17 are all made up of known members, mechanisms, devices, etc., and detailed illustrations and explanations of each component will be omitted.

[0013] The detection device 12 is composed of a computer and various devices for executing detection processing of various information related to objects including collaborators, walls, and other obstacles present around the robot 10. The detection device 12 includes a position detection unit 19 for detecting object position information, a speed detection unit 20 for detecting object speed information based on the detection result by the position detection unit 19, and a voice detection unit 21 such as a microphone for detecting voice information from collaborators, and is capable of acquiring each of these pieces of information at predetermined time intervals.

[0014] The position detection unit 19 is not particularly limited, but is composed of known sensors including a range sensor such as a known laser range finder, and a depth sensor that uses camera information. These sensors acquire the position coordinates of the object at the detection time, and determine whether it is the same object or not based on the time series data. For moving objects such as the same collaborator, the speed detection unit 20 obtains speed information from the change in the acquired time series position information.

[0015] In the speed detection unit 20, the speed information of the collaborators around the robot 10 is calculated from the position information by a method already proposed by the present inventors (JP Patent Publication No. 2020-91235) or the like every time the position information of the collaborators around the robot 10 is acquired. The algorithm for deriving the speed information is not an essential element of the present invention, so a detailed description thereof will be omitted.

[0016] As the detection device 12, in addition to the above-mentioned embodiments, various sensors, devices, systems, etc. can be applied as long as they are capable of detecting the above-mentioned various information.

[0017] As described above, information regarding objects present within a specified range around the robot 10 is detected sequentially by the detection device 12 at specified time intervals, and is transmitted to and stored in the control device 13 at the detection timing, and the following processing is performed by the control device 13 at each detection timing.

[0018] The control device 13 is provided integrally with the robot 10 or separately, and is configured by a computer including a processor such as a CPU and storage devices such as a memory and a hard disk. Programs for causing the computer to function as each of the following units are installed in the computer.

[0019] The control device 13 includes a memory unit 23 in which various data are stored, a collaborator behavior estimation unit 24 that estimates the behavior of a collaborator with respect to a task based on changes over time in the position information and speed information of the collaborator, a motion plan setting unit 25 that sets a motion plan including a movement trajectory of the robot 10 in accordance with the behavior of the collaborator estimated by the collaborator behavior estimation unit 24, and an operation command unit 26 that controls the operation unit 11 so that the robot 10 operates in accordance with the motion plan set by the motion plan setting unit 25.

[0020] The memory unit 23 accumulates various information successively detected by the detection device 12, and also stores preset task information, relational expressions and fixed parameters for various calculations, and data of the following processing results.

[0021] Here, as the task information, when the robot 10 and the collaborator perform a series of tasks set in advance, the robot 10 is assigned m tasks T R {T1 R ,T2 R ,···,T m R}, and each collaborator is given n tasks T H {T1 H ,T2 H ,···,T n H For each task, in addition to the target position and the operation content (grab, place, etc.), a robot target work time t avg R (T R ), and the collaborator's estimated work time t avg H (T HIn addition, the conditions of the previous and next process tasks that indicate the context of the tasks (e.g., the execution of a grab action is required to execute a place action) are set in advance.

[0022] The collaborator behavior estimation unit 24 includes an achievement level estimation unit 28 that estimates the achievement level of each task of the collaborator, a collaborator action planning unit 29 that predicts a movement trajectory when the collaborator performs each task and calculates energy corresponding to the ease of movement to each task based on the predicted movement trajectory, a likelihood calculation unit 30 that calculates the likelihood of the achievement level and the energy from the processing results of the achievement level estimation unit 28 and the collaborator action planning unit 29, and a Bayesian update unit 31 that periodically updates the task execution probability, which is the probability that the collaborator will next execute each task, by Bayesian estimation based on the likelihood calculated by the likelihood calculation unit 30.

[0023] The achievement level estimation section 28 performs the following calculation process.

[0024] In calculating the probability of a collaborator performing a task, the task completion level, which indicates the degree to which a collaborator has already accomplished each task, is important information. This task completion level A(T) is defined as a continuous value that has an initial value of 0 and becomes 1 when the task is completely accomplished, and is set so that the higher the task completion level, the lower the probability of performing the task.

[0025] The achievement level estimation unit 28 estimates the achievement level of each of the n tasks T H {T1 H ,T2 H ,···,T n H Task achievement level A(T H ){A(T1 H ),A(T2 H ),····,A(T n H )} is estimated as follows:

[0026] First, the periodically acquired collaborator position x H, velocity v H Then, the collaborator performs task T H {T1 H ,T2 H ,···,T n H} within the feasible distance range d max (T) is the number of seconds t that the device is still in elap (T) is calculated. The position of the collaborator x H and the task T H {T1 H ,T2 H ,···,T n H} is the distance difference between the target positions Δd and the minimum speed at which a human is considered to be stationary is v min H Then, the number of seconds of stillness t elap (T) is calculated using the following formula:

number

number

[0027] The collaborator motion planning unit 29 performs the following calculation process.

[0028] In estimating the probability of a collaborator performing a task, important information includes whether the collaborator tends to move toward the target position of the task and how easy it is to reach the target position. To obtain this information, it is necessary to predict the movement trajectory of the collaborator when moving toward the task. At this time, taking into consideration situations such as the collaborator trying to make a detour or avoidance, or there being no passable road, the movement trajectory when moving toward each task is calculated using the following method. Here, the method (Adaptive Goal Finder (hereinafter referred to as "AGF")) already proposed by the present inventors in JP 2023-32381 A is applied to the task T of the collaborator from the collaborator's perspective rather than the robot's perspective. H {T1 H ,T2 H ,···,T n H} and for each task T H {T1 H ,T2 H ,···,T n H The AGF is a method for searching for a target position based on sequential prediction of the work content and the interaction between humans and the robot in the environment. The specific algorithm is not essential to the present invention, so a detailed description will be omitted.

[0029] Next, for each task T H {T1 H ,T2 H ,···,T n H} for each task T H {T1 H ,T2 H ,···,T n H}, if a collaborator H {T1 H ,T2 H ,···,T n H}'s tendency to move toward the destination (speed directional difference) and the energy E(T) (E(T H ){E(T1 H ),E(T2 H),····,E(T n H )}) is calculated.

[0030] The speed direction difference θ diff (T) is the velocity vector v of the collaborator as follows: H For each task T H {T1 H ,T2 H ,···,T n H}'s moving trajectory's direction vector v T H (T) is calculated by the angle difference of θ diff (T)=1.

number

[0031] The orbit cost c path (T) is described in JP 2023-32381 A, and the cost of the movement trajectory obtained during processing in the AGF is used as is.

[0032] The above speed difference θ diff (T) and orbital cost c path From (T), for each task T H {T1 H ,T2 H ,···,T n H}, energy E(T) serving as a parameter used in likelihood calculation section 30 is calculated by the following equation.

number

[0033] The likelihood calculation unit 30 calculates the task achievement degree A(T H ), the energy E(TH ) the likelihood of each task P(A,E|T H ){P(A(T1 H ),E(T1 H )|T1 H ),P(A(T2 H ),E(T2 H )|T2 H ),···,P(A(T n H ),E(T n H )|T n H In calculating the likelihood, the random variable A(T H ), E(T H ) Consider a single truncated normal distribution with ≤ 1.

number

[0034] The Bayes update unit 31 updates the likelihood (A,E|T H ) to obtain the Bayesian estimation of the task execution probability P(T H ) is updated. At this time, the collaborators update all tasks T H {T1 H ,T2 H ,···,T n H} and which task T H {T1 H ,T2 H ,···,T n H Consider the probability P(φ) that none of the steps have been executed. P(φ) is calculated using the following formula:

number

[0035] P(T H |A,E){P(T1 H |A(T1 H ),E(T1 H)),P(T2 H |A(T2 H ),E(T2 H )),···,P(T n H |A(T n H ),E(T n H The posterior probability represented by {\displaystyle \mathbb {P\text{R}\text{R}}} is calculated by the following formula:

number

[0036] In order to quantitatively evaluate and select the planned behavior, the motion plan setting unit 25 calculates a collaborative work cost corresponding to the load related to the collaborative work between the robot 10 and a collaborator for each of the robot 10's preset behavior options, and selects the behavior of the robot 10 based on the collaborative work cost.

[0037] Here, the behavioral options include a normal behavior, which is a behavioral state in which the robot 10 and the collaborator execute tasks given to each other in advance; a request behavior, which is a behavioral state in which the robot 10 asks the collaborator to take over at least a part of its task; and a support behavior, which is a behavioral state in which the robot 10 takes over at least a part of the collaborator's task.

[0038] This motion plan setting unit 25 includes a cost calculation unit 33 that calculates the collaborative work cost for each of the collaborators and the robot 10 in each behavior option, and a behavior selection unit 34 that selects an appropriate behavior from the behavior options based on the collaborative work cost calculated by the cost calculation unit 33.

[0039] The cost calculation unit 33 determines the collaborative work cost by separately calculating and adding up the cost of the task currently being executed (hereinafter referred to as the "immediate task") and the cost of other tasks (hereinafter referred to as the "remaining tasks").

[0040] Here, the cost of the worker's most recent task T0 is C(T0), and the remaining tasks T i The cost of C(T i ) is calculated using the following formula:

number

[0041] On the other hand, the cost C(T0) of the nearest task T0 of the robot 10 is determined as the sum of the movement time required to perform the nearest task T0 and the estimated work time for the nearest task T0 stored in advance, based on the movement trajectory of the robot 10 generated by the AGF, which is the above-mentioned behavior planning method, and the movement speed of the robot 10 that is set in advance, for each behavior option of the robot 10. i The cost of C(T i ) is the remaining task T i Since the order of the remaining tasks is clear, i and the distance between the task and the previous task is l i Let the distance l be i The moving time is calculated by dividing the moving speed of the robot 10 by the moving speed of the robot 10 set in advance, and the remaining task T i This is the sum of the estimated working hours in

[0042] And the collaborative work cost of robot 10 and the collaborator C R , C H is calculated from the following formula:

number

[0043] In addition, in the case of normal actions, the collaborator's immediate task T0 is task T H {T1 H ,T2 H ,···,T n H}, the task execution probability P(T H ) is considered to be the case when performing the highest task.

[0044] In the case of a request action, the immediate task T0 of the robot 10 approaches the collaborator and hands over the task, and the immediate task T0 of the collaborator receives the task.

[0045] Furthermore, in the case of a supportive action, the immediate task T0 of the robot 10 approaches the collaborator and receives the task of the collaborator, and the immediate task T0 of the collaborator hands over the task.

[0046] In addition, the cost calculation unit 33 plans the motion of the robot 10 for each action option by the AGF. Here, when planning a request motion or a support motion, global path planning is executed between the robot 10 and the collaborator, and a point on the trajectory that is equidistant from both is set as a virtual target position. However, after the robot 10 and the collaborator get sufficiently close, the position of the collaborator is set as the target position, and motion planning is performed by changing the grid granularity in the AGF and the maximum workable distance.

[0047] In the following description, in the case of normal behavior in which both parties execute a task given to each other, the collaborative work cost of the robot 10 is C R C Nrml R , collaboration cost of collaborators C H C Nrml H In addition, in the case of a request action in which the robot 10 requests a task from a collaborator, the collaborative work cost of the robot 10 is C R C Rqst R , collaboration cost of collaborators C H C Rqst H Furthermore, in the case of a support behavior in which the robot 10 supports a task of a collaborator, the collaboration task cost of the robot 10 is C R C Asst R , collaboration cost of collaborators C H C Asst HBoth are calculated using the above-mentioned formula, but when the robot 10 performs a request action or a support action, the cost of the most recent task of the robot 10 is calculated as a task handover action, and the cost of the most recent task of the collaborator is set to be equal to the cost of the most recent task of the robot 10.

[0048] The action selection unit 34 compares each of the normal action, the request action, and the support action using a conditional formula described later based on a total cost that combines the collaborative work costs of the robot 10 and the cooperator for each of these actions, and determines the action of the robot 10 based on the result. At this time, the total cost for each of the normal action, the request action, and the support action is calculated by multiplying the collaborative work cost C R , C H The total value of the preset coefficient K Nrml , K Rqst , K Asst These coefficients K Nrml , K Rqst , K Asst The default value is 1 and it changes depending on the following conditions (1) to (3).

[0049] (1) Reactive constants In the requesting action and the supporting action of the robot 10, a proactive state is defined in which the robot asserts itself to the collaborator, and a reactive state in which the robot 10 conforms to the assertion of the collaborator. These are simply determined by which of the robot 10 or the collaborator first proposes a collaborative compromise, such as a requesting action or a supporting action. Therefore, if the collaborator proposes a collaborative compromise to the robot 10 first, the robot 10 will decide to act reactively. This determination is made by uttering "please" or "I'll do it" to the other party, and determining that the one who made the utterance has proposed a requesting action or a supporting action to the other party. When the robot 10 decides to act reactively, it is necessary to take into consideration as a cost element the fact that the cooperator has proposed a cooperative compromise. Therefore, when the voice detection unit 21 detects speech information of the cooperator saying "I'll do it" or a synonymous utterance, and the robot 10 is determined to be in a reactive state, the coefficient KRqst , the reactive coefficient K R On the other hand, when the speech information of the collaborator "please" or a synonym thereof is detected and the robot 10 is determined to be in a reactive state, the coefficient K Asst , the reactive coefficient K R These reactive coefficients K R is a fixed parameter stored in the storage unit 23, and setting it to less than 1 makes the behavior more likely to occur. (2) Behavioral change constant When the robot 10 selects a request action or a support action and proactively expresses its intention to the collaborator by speaking to the speaker 16, selecting another action immediately after that may confuse the collaborator. Therefore, once a collaborative compromise action is selected, the coefficient of the other action is not changed by the action change constant K C This behavior change constant K C is a fixed parameter stored in the storage unit 23, and by setting it to a value greater than 1, the behavior in question becomes less likely to occur. (3) Prediction error variables If the cooperator fails to respond for some reason after the robot 10 starts the collaborative yielding, the robot 10 must abandon the collaborative yielding and change its behavior. Therefore, it is determined whether the behavior calculated at the time the collaborative yielding started is progressing as predicted, and a prediction error variable K E is multiplied by the coefficient of the action in question. E is calculated by the following formula:

number

[0050] The behavior of the robot 10 is determined by the following conditional expressions:

number

[0051] In the above formula, the request action is selected when the total cost obtained by multiplying the sum of the collaborative work costs of both the robot 10 and the collaborator by the coefficient is smaller than that of the normal action, and the collaborative work cost of the robot 10 is larger than that of the collaborator. The support action is selected when the total cost is smaller than that of the normal action, and the collaborative work cost of the robot 10 is smaller than that of the collaborator. Here, in the case of reactive behavior, the action is selected only by comparing the total cost with the normal action, without comparing the collaborative work costs between the robot 10 and the collaborator. Then, when neither of these conditions is satisfied, the normal action is selected. In the above, the total cost of the collaborative work costs of both the robot 10 and the collaborator is compared between the collaborative yielding action and the normal action, and the action that reduces the total amount of work of the robot 10 and the collaborator is selected.

[0052] In addition, in order to select an action that shortens the total task time, the action of the robot 10 can be determined by adopting other conditional expressions instead of the above-mentioned conditional expressions.

[0053] The motion command unit 26 issues a motion command to the motion unit 11 so as to perform a task execution motion including movement on a movement trajectory in accordance with the action of the robot 10 selected by the motion plan setting unit 25 .

[0054] Here, when a request action or a support action is selected, the robot 10 performs a speaking action so that the intention is conveyed to the collaborator through a voice from the speaker 16. Also, when the voice detection unit 21 detects speech information indicating the intention of the collaborator to decline the assertion of the robot 10, the robot 10 can switch to normal action by re-planning the action. Conversely, when the robot 10 selects normal action when the collaborator makes an assertion, this is an action of declining the assertion of the collaborator, so the robot performs a speaking action so that the intention is conveyed to the collaborator through a voice from the speaker 16.

[0055] In addition, when a task is handed over due to a request action and a support action of the robot 10, the task T of the robot 10 and the collaborator stored in the storage unit 23 is R , T H In the case of a request action, the task to be requested is updated to the task T R is deleted from the collaborator's task T H In this case, the collaborator starts executing the task on the spot, so the execution probability of the task is set to 1 or a value slightly smaller than 1. In the case of a support action, the task T H is deleted, and task T of robot 10 is R At this time, the probability of executing other tasks of the collaborators is recalculated so that the relative percentage to the current total value does not change.

[0056] The control device 13 configured as described above functions as a behavior decision system that decides the behavior of the robot 10 that executes a series of operations consisting of a plurality of predetermined tasks in cooperation with a collaborator.

[0057] The series of operations targeted by the present invention is not particularly limited, and the following operations and tasks can be exemplified. In the embodiment, the delivery of an object between the robot 10 and a collaborator is also assumed as a proactive or reactive behavior. In this case, it is also possible to consider an action option other than the delivery, such as moving a trash can or the like that is the destination point of the object, taking into account work efficiency. It is also possible to set a subgoal located between the robot 10 or collaborator and the destination point as an action option other than the delivery. In addition, the present invention can be applied to various other operations, such as cooking while the robot 10 and a collaborator work together.

[0058] In the above embodiment, the collaborative work cost is calculated based on the required time, but the present invention is not limited to this. The collaborative work cost can also be calculated based on the distance or the amount of work. In this case, the collaborative work cost can be calculated by changing the time parameter in the above-mentioned cost formula to a distance or amount of work parameter. Here, an example of the amount of work parameter is a numerical value obtained by multiplying the weight of the transported luggage by the amount of movement.

[0059] Additionally, the configuration of each part of the device in the present invention is not limited to the illustrated configuration example, and various modifications are possible as long as they provide substantially the same effect. [Explanation of symbols]

[0060] 10. Robot 11 Operating unit 12 Detection device 13 Control device 24 Collaborator Behavior Estimation Department 25 Motion planning unit 26 Operation command section 28 Achievement level estimation section 29 Collaborator Motion Planning Section 30 Likelihood calculation section 31 Bayesian Update Section 33 Cost Calculation Department 34 Action Selection Section

Claims

1. A robot comprising: an operating unit that operates to be able to execute predetermined tasks; a detection device that detects surrounding environmental information; and a control device that controls the operation of the operating unit based on a detection result of the detection device, the robot executing a series of operations consisting of a plurality of the tasks in cooperation with a collaborator, The detection device detects position information and velocity information of objects around the robot, including a collaborator; the control device includes a collaborator behavior estimation unit that estimates a behavior of the collaborator with respect to the task based on time-dependent changes in the position information and the speed information of the collaborator, an action plan setting unit that sets an action plan including a movement trajectory of the robot according to the action of the collaborator estimated by the collaborator behavior estimation unit, and an action command unit that controls the action unit so that the robot operates in accordance with the action plan, The motion plan setting unit creates motion plans for each of the action options consisting of a normal action in which the robot and the collaborator execute the task that has been given to each other in advance, a request action in which the robot asks the collaborator to take over at least a part of the task, and a support action in which the robot takes over at least a part of the task of the collaborator, calculates a collaborative work cost corresponding to the load of the task to be executed by the robot and the collaborator, compares each of the actions in the action options based on the collaborative work cost, and selects the motion plan for the appropriate action.

2. A robot comprising: an operating unit that operates to be able to execute predetermined tasks; a detection device that detects surrounding environmental information; and a control device that controls the operation of the operating unit based on a detection result of the detection device, the robot executing a series of operations consisting of a plurality of the tasks in cooperation with a collaborator, The detection device detects position information and velocity information of objects around the robot, including a collaborator; the control device includes a collaborator behavior estimation unit that estimates a behavior of the collaborator with respect to the task based on time-dependent changes in the position information and the speed information of the collaborator, an action plan setting unit that sets an action plan including a movement trajectory of the robot according to the action of the collaborator estimated by the collaborator behavior estimation unit, and an action command unit that controls the action unit so that the robot operates in accordance with the action plan, the operation plan setting unit includes a cost calculation unit that calculates a collaborative work cost corresponding to a sum of a load of a currently executed immediate task and a load of other remaining tasks for each of the collaborator and the robot, and a behavior selection unit that selects an appropriate behavior from preset behavior options based on the collaborative work cost; As the action options, a normal action in which the robot and the collaborator execute the task given to each other in advance, a request action in which the robot requests the collaborator to substitute for at least a part of the task of the robot, and a support action in which the robot substitutes for at least a part of the task of the collaborator are set, The robot is characterized in that the cost calculation unit calculates the collaborative task cost for each of the action options.

3. 3. The robot according to claim 2, wherein the action selection unit compares the requested action or the supporting action with the normal action based on a total cost which is an aggregate cost of the collaborative work of the collaborator and the robot, and selects the requested action or the supporting action on the condition that the requested action or the supporting action is smaller than the normal action.

4. The robot according to claim 2, characterized in that, when the behavior selection unit determines that the robot is in a reactive state in which it is in tune with the request or support behavior of the collaborator, a predetermined coefficient is multiplied by the collaborative task cost, so as to make the robot more likely to perform that action.

5. 3. The robot according to claim 2, wherein the behavior selection unit is configured such that, when the robot selects the request behavior or the support behavior to put the robot in a proactive state in which the robot asserts the request behavior or the support behavior to a collaborator, a predetermined coefficient is multiplied against the collaborative task cost of another behavior until the most recent behavior is completed, making it difficult to perform the other behavior.

6. 3. The robot according to claim 2, wherein the action selection unit multiplies the collaborative task cost by a predetermined coefficient so that, if the cooperator is unable to respond for some reason after the robot starts the request action or the support action, the action is abandoned and changed to another action.

7. 3. The robot according to claim 2, wherein the collaborator behavior estimating unit comprises an achievement level estimating unit that estimates an achievement level of each of the tasks by the collaborator, a collaborator action planning unit that estimates a movement trajectory when the collaborator performs each of the tasks and calculates energy corresponding to the ease of moving to each of the tasks based on the movement trajectory, a likelihood calculating unit that calculates a likelihood of the achievement level and the energy, and a Bayesian updating unit that periodically updates a task execution probability, which is the probability that the collaborator will perform each of the tasks, by Bayesian estimation from the likelihood.

8. A system for determining actions of a robot that executes a series of operations consisting of a plurality of predetermined tasks in cooperation with a collaborator, comprising: a collaborator behavior estimation unit that estimates a behavior of the collaborator with respect to the task based on time-dependent changes in position information and speed information of the collaborator, and an action plan setting unit that sets an action plan including a movement trajectory of the robot according to the action of the collaborator estimated by the collaborator behavior estimation unit, The action plan setting unit creates the action plans for each of the action options consisting of a normal action in which the robot and the collaborator execute the task that has been given to each other in advance, a request action in which the robot asks the collaborator to take over at least a part of the task, and a support action in which the robot takes over at least a part of the task of the collaborator, calculates a collaborative work cost corresponding to the load of the task to be executed by the robot and the collaborator, compares each of the actions in the action options based on the collaborative work cost, and selects the action plan for the appropriate action, characterized in that

9. A program for a system that determines the behavior of a robot that executes a series of operations consisting of a plurality of predetermined tasks in cooperation with a collaborator, a collaborator behavior estimation unit that estimates a behavior of the collaborator with respect to the task based on time-dependent changes in position information and speed information of the collaborator, and an action plan setting unit that sets an action plan including a movement trajectory of the robot according to the action of the collaborator estimated by the collaborator behavior estimation unit; The action plan setting unit creates the action plans for each of the action options consisting of a normal action in which the robot and the collaborator execute the task that has been given to each other in advance, a request action in which the robot asks the collaborator to take over at least a part of the task, and a support action in which the robot takes over at least a part of the task of the collaborator, calculates a collaborative work cost corresponding to the load of the task to be executed by the robot and the collaborator, compares each of the actions in the action options based on the collaborative work cost, and selects the action plan for the appropriate action, characterized in that

Citation Information

Patent Citations

  • Robot, action planning device, and action planning program

    JP2020046773A

  • Robot, optimum goal search device and its program

    JP2023032381A