Action Transfer Device Using Partial Destination Samples

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current action transfer techniques require a large number of learning samples from both the transfer source and destination robots, making the process time-consuming and costly, especially when the transfer destination robot is a real robot in an actual operation setting.

Innovation Solution

An action transfer method that modifies a large number of data sets from the transfer source robot using a small number of action samples from the transfer destination robot, optimizing the data sets to enable efficient action transfer without the need for a large number of learning samples at the destination.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If a large number of learning samples are acquired from both transfer source and destination robots using conventional action transfer techniques, then the action transfer accuracy is improved, but the time consumption and cost increase significantly

Engineering Contradiction:
Improveaction transfer accuracyVSAvoidtime consumption
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The patent applies partial action by acquiring only a small number of learning samples from the transfer destination robot (e.g., 10-100 samples) instead of requiring equal large numbers from both source and destination robots. The correction unit then processes these partial samples to achieve accurate action transfer, resolving the contradiction between sample quantity and time consumption

Inventive Principle:
Principle #16Partial or excessive action

Solution Approach 2:

The patent performs preliminary action by acquiring action information from the transfer source robot in advance and storing it as a dataset. This pre-acquired information is then corrected using a small number of destination robot samples, eliminating the need to collect large numbers of samples from the destination robot at transfer time, thus reducing time consumption

Inventive Principle:
Principle #10Preliminary action

2Measurement precision

If a large number of learning samples are acquired from both transfer source and destination robots, then the action transfer accuracy is improved, but the man-hours and costs increase

Engineering Contradiction:
Improveaction transfer accuracyVSAvoidman-hours and cost efficiency
Core Design Contradiction:
Measurement precisionVSProductivity

Solution Approach 1:

The patent requires only partial action in terms of sample acquisition from the destination robot (a small number of samples), while utilizing the pre-acquired action information from the source robot. This dramatically reduces the manual effort and cost associated with collecting equal numbers of samples from both robots, improving productivity

Inventive Principle:
Principle #16Partial or excessive action

Solution Approach 2:

The patent copies action information from the transfer source robot to the transfer destination robot through the correction unit. Instead of requiring the destination robot to independently learn through extensive sampling, the system copies and adapts the source robot's action information using minimal destination samples, reducing man-hours and costs

Inventive Principle:
Principle #26Copying

3Measurement precision

If physical properties of the transfer destination robot are obtained in advance through measurement, then the action adaptation accuracy is improved, but the complexity of advance preparation increases

Engineering Contradiction:
Improveaction adaptation accuracyVSAvoidcomplexity of advance preparation
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent introduces an intermediary approach by using action information from the transfer source robot as a mediator between the source and destination robots. Instead of directly measuring and adapting to the destination robot's physical properties (which requires complex preparation), the system uses the source robot's action information as an intermediate reference that can be corrected with minimal destination samples

Inventive Principle:
Principle #24Intermediary (Mediator)

Solution Approach 2:

The patent changes the approach from directly measuring physical properties to processing action information parameters. The correction unit modifies the action information parameters (joint angles, end effector coordinates) based on a small number of destination samples, avoiding the need for complex physical property measurement and adaptation

Inventive Principle:
Principle #35Parameter changes

Data Source

PatentEP3228427B1Action transfer device, action transfer method, and non-transitory computer-readable medium in which program is stored
Publication Date: 2020.07.22 SOINN INC
  • EP3228427B1 patent drawingFigure 1
  • EP3228427B1 patent drawingFigure 2
  • EP3228427B1 patent drawingFigure 3

AI summary

Provided are an action transfer device, an action transfer method, and a program which are capable of transferring an action from a transfer source robot using a small number of learning samples acquired by a transfer destination robot. In an action transfer device (100), a transfer source operation information acquisition unit (101) acquires a plurality of pieces of action information about the transfer source robot; a transfer destination operation information acquisition unit (102) acquires a plurality of pieces of first action information about the transfer destination robot; and a correction unit (103) generates a plurality of pieces of second action information about the transfer destination robot by correcting action information about the transfer source robot by a prescribed update formula using the first action information about the transfer destination robot. The number of the pieces of the first action information about the transfer destination robot is smaller than the number of the pieces of the action information about the transfer source robot, and the number of the pieces of the second action information about the transfer destination robot is larger than the number of the pieces of the first action information about the transfer destination robot.