Robot Action Transfer Using Few-Shot Error Propagation
Find Innovative SolutionsGenerate 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, and it is difficult to achieve rapid and accurate action transfer, especially when the physical properties of the transfer destination robot are unknown.
Innovation Solution
An action transfer device and method that uses a small number of learning samples from the transfer destination robot to correct and adapt action information from a transfer source robot, employing an update formula to generate transferable action data, which includes joint values and coordinate values, and repeatedly propagates errors until a preset threshold is met, allowing for efficient action transfer without requiring extensive data sets from the transfer destination robot.
Engineering Contradictions & Design Principles
Engineering 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 processing time and cost increase significantly
Solution Approach 1:
The patent extracts only the essential correction information from a small number of destination robot samples (coordinate values and joint values) and uses this extracted information to correct the action information from the source robot through error propagation, rather than requiring large numbers of samples from both robots
Solution Approach 2:
The patent introduces an error propagation mechanism as an intermediary that transfers correction information from the small set of destination samples to the source robot's action information, enabling accurate action transfer without directly matching large datasets
2Measurement precision
If conventional action transfer techniques are used requiring large datasets from both robots, then the action transfer accuracy is improved, but the number of man-hours and cost increase
Solution Approach 1:
The patent applies partial action by using only a small number of samples from the destination robot (exactly what is needed for correction) rather than requiring equal or excessive numbers of samples from both robots, reducing implementation cost while maintaining accuracy
3Measurement precision
If physical properties of the transfer destination robot are obtained in advance for action transfer, then the action transfer accuracy is improved, but the complexity of advance preparation increases
Solution Approach 1:
The destination robot performs self-service by autonomously executing the corrected action information generated by the correction unit, allowing the system to adapt to unknown physical properties without complex advance preparation or measurement procedures
Data Source
AI summary
A transfer source action information acquisition unit acquires first action information of the transfer source robot. A transfer destination action information acquisition unit acquires second action information of the transfer destination robot. A correction unit corrects the action information of the transfer source robot by using the second action information and in accordance with a predetermined update formula and thereby generates third action information of the transfer destination robot. The pieces of the second action information of the transfer destination robot are less than the pieces of the first action information of the transfer source robot. The first to third action information includes a set of data indicative of one or more robot joint values and a set of data indicative of a coordinate value of a robot specific part.