Humanoid Robot Gripper Control for Skilled Cloth Handling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional robots struggle to perform fine motion tasks, such as sewing, due to the difficulty in replicating the intricate movements of a skilled person's hand and arm, especially when handling workpieces like cloth.
Innovation Solution
A humanoid robot equipped with a gripping portion comprising a palm portion and multiple finger portions, a palm sensor unit, and a control unit that utilizes learning data and a trained model to mimic the operations of a skilled worker, adjusting grip and motion based on detected workpiece information and machine tool operations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If a robot uses a finger-type gripping portion to perform fine motion work, then the robot can handle workpieces like cloth, but it is difficult to reproduce the motion of a skilled person's hand and arm
Solution Approach 1:
The patent applies the copying principle by using a trained model that replicates the operation patterns of a skilled worker. The learning data generation unit creates training data from motion capture information of skilled workers, and the trained model copies these human operation patterns to control the robot's gripping portion, enabling the robot to reproduce skilled human motions accurately
Solution Approach 2:
The patent implements feedback through the learning data generation unit that captures actual motion data from skilled workers and feeds it back into the trained model. This continuous feedback loop allows the system to refine and improve the robot's motion reproduction accuracy by comparing predicted motions with actual skilled worker motions
2Ease of manufacture
If the robot structure is simplified, then the robot is easier to manufacture, but it cannot reproduce complex skilled hand and arm motions
Solution Approach 1:
The patent replaces complex mechanical control systems with an AI-based trained model. Instead of using complex mechanical structures to achieve skilled motions, the system uses a learned model that processes sensor data and generates appropriate motion commands, substituting mechanical complexity with intelligent control
Solution Approach 2:
The patent changes the control parameters from fixed mechanical control to dynamic learned parameters. The trained model adjusts motion parameters based on learned patterns from skilled workers, allowing the robot to perform complex tasks without corresponding mechanical complexity
3Device complexity
If the robot uses conventional control methods, then the control system is simpler, but it cannot achieve fine motion control like a skilled worker
Solution Approach 1:
The control system copies skilled worker behavior patterns through the trained model. The learning data generation unit captures and stores the operational patterns of skilled workers, and the trained model reproduces these patterns to control the robot, achieving fine motion control without requiring complex conventional control mechanisms
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A humanoid robot (robot) includes: a gripping portion that includes a palm portion serving as a base for holding cloth (workpiece) and plural finger portions radially extended from the palm portion; a palm sensor unit that is provided at the palm portion and detects workpiece information including a shape and a disposition of the workpiece; and a control unit that controls holding and an operation of the workpiece by the gripping portion based on an operation parameter obtained by inputting a detection result of the palm sensor unit into a trained model learned using learning data representing a combination of an operation of a hand of a skilled worker during work and an operation of a sewing machine and executing arithmetic processing of the learning model.