Robot Action Learning for Human Burden and Efficiency Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional robot systems that allow human-robot collaboration face challenges in optimizing robot control methods to minimize human burden and maximize efficiency, as various human action patterns and task methods complicate the setting of appropriate control strategies.
Innovation Solution
A machine learning device that observes state variables of the robot and human collaboration, obtains determination data on human burden and efficiency, and learns a training data set to optimize robot actions using reinforcement learning, with components like state observation, determination data acquisition, reward computation, and action value updates.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the robot trajectory is set in advance to drive along a predetermined path, then the robot operation is simplified and reliable, but the adaptability to different human action patterns and task requirements is reduced
Solution Approach 1:
The patent applies dynamics by transitioning from static predetermined trajectories to dynamic adaptive trajectories. The robot controller continuously adjusts the robot's motion trajectory based on real-time observation of human action patterns, allowing the system to adapt to different human operators and task requirements while maintaining reliable operation through structured learning frameworks
2Adaptability or versatility
If numerous control methods are available for robot manipulation, then the system flexibility increases, but the difficulty of selecting and setting the optimal control method increases
Solution Approach 1:
The patent implements feedback mechanisms where the robot controller observes human action patterns and determines burden levels and working efficiency. This feedback loop enables the system to automatically evaluate and select optimal control methods based on real-time performance metrics, reducing the complexity of control setting while maintaining high flexibility through data-driven decision making
3Productivity
If the robot works cooperatively with human without safety fence, then the working efficiency increases, but the safety risk to human increases
Solution Approach 1:
The patent replaces mechanical safety fences with an intelligent control system that uses sensors and machine learning algorithms to monitor human-robot collaboration. This substitution allows uninterrupted cooperative work while maintaining safety through real-time observation of human action patterns and automatic adjustment of robot behavior to prevent harmful situations
4Productivity
If the robot learns and adapts to individual human action patterns, then the working efficiency and comfort increase, but the data processing complexity and learning time increase
Solution Approach 1:
The patent applies preliminary action by pre-processing and structuring human action pattern data during the learning phase. The robot controller collects and organizes observation data about human actions, burden levels, and efficiency metrics in advance, creating structured training datasets that facilitate faster and more efficient learning while reducing real-time data processing complexity
Data Source
AI summary
A machine learning device for a robot that allows a human and the robot to work cooperatively, the machine learning device including a state observation unit that observes a state variable representing a state of the robot during a period in that the human and the robot work cooperatively; a determination data obtaining unit that obtains determination data for at least one of a level of burden on the human and a working efficiency; and a learning unit that learns a training data set for setting an action of the robot, based on the state variable and the determination data.