Wheel-Legged Robot Control Using Data-Driven Equilibrium Modeling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Controlling the motion of under-actuated robots, particularly wheel-legged robots, is challenging due to the complexity of their mechanical structure, making it difficult to obtain an accurate dynamic model and perform parameter identification, which affects the controller's effectiveness.
Innovation Solution
A method using adaptive dynamic programming and value iteration to calculate a controller that adapts to the robot's dynamic characteristics, allowing it to maintain equilibrium and follow a target trajectory, even with unknown dynamic parameters, by collecting and processing motion and control data to build a linear equilibrium parameter matrix.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If an accurate dynamic model is designed for the robot, then the controller effectiveness is improved, but the device complexity increases due to the complex mechanical structure
Solution Approach 1:
The patent replaces the traditional mechanical modeling approach with a data-driven iterative learning approach. Instead of deriving complex dynamic models from the mechanical structure, the system collects motion state data and control data during robot operation, then iteratively identifies dynamic parameters through data processing, substituting the mechanical analysis system with an information processing system.
Solution Approach 2:
The robot system performs self-identification of its dynamic parameters through iterative learning using its own operational data. The system collects data from its own motion processes and uses this data to automatically identify its dynamic characteristics without requiring external modeling assistance, enabling the system to serve its own modeling needs.
2Measurement precision
If parameter identification is performed in the dynamic model, then the controller accuracy is improved, but the loss of time increases due to the difficulty of identification
Solution Approach 1:
The patent performs preliminary data collection during the robot's normal motion processes before the actual parameter identification is needed. By accumulating motion state data and control data during routine operations, the system prepares the necessary identification data in advance, reducing the time required when actual parameter identification is performed.
Solution Approach 2:
The data collection process continues throughout the robot's operational life rather than being a discrete event. The system continuously collects and accumulates motion data during normal operations, transforming the parameter identification process from a time-consuming separate task into a continuous background process that occurs alongside useful robot operations.
3Measurement precision
If the robot uses a controller based on known dynamic parameters, then the motion control accuracy is improved, but the adaptability decreases when parameters are not accurate
Solution Approach 1:
The patent transforms the static controller into a dynamic adaptive controller that continuously updates its understanding of the robot's dynamic parameters. The controller evolves from a fixed-parameter system to a time-varying system that adapts its control strategy based on iteratively identified parameters from actual operational data, enabling it to adjust to parameter uncertainties and changes.
Solution Approach 2:
The system implements a feedback loop where the robot's actual motion performance is continuously monitored, and this information is fed back into the parameter identification process. The identified parameters are then fed back into the controller to improve its accuracy, creating a closed-loop system that continuously refines its control based on actual performance data.
Data Source
AI summary
A method for configuring a controller for a wheel-legged robot includes: controlling motion of the robot, and obtaining motion state data and control data of the robot during a motion process, where diversity measures of the motion state data and the control data are higher than a predetermined threshold; calculating a linear equilibrium parameter matrix by using a data iteration method according to the motion state data and the control data; and configuring a controller corresponding to dynamic characteristics of the robot based on the linear equilibrium parameter matrix.