Legged Robot Kinematics Neural Network Real-Time Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The high computational complexity of forward kinematics in biped robots makes real-time control challenging due to the difficulty in solving high-order equations, which numerical methods based on the Jacobian matrix and Newton-Raphson method cannot efficiently address.
Innovation Solution
Transforming the complex forward kinematics solving process into a simpler inverse kinematics solving process by training a neural network model using a sufficient number of training samples, where the output of inverse kinematics is used as input and vice versa, allowing the neural network to process forward kinematics solving.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If numerical methods based on Jacobian matrix and Newton-Raphson method are used to solve forward kinematics, then the robot can perform motion trajectory planning, but the computational complexity is high and processing time is long
Solution Approach 1:
The patent pre-calculates and stores a mapping relationship between joint angles and end effector positions in advance, creating a lookup table that can be quickly queried during real-time control. This preliminary computation phase separates the complex calculation from the time-critical control phase, resolving the contradiction between accuracy and speed.
Solution Approach 2:
The patent creates a simplified computational model (lookup table) that copies the essential input-output relationship of the complex forward kinematics system. Instead of performing full numerical calculations during control, the system queries pre-computed values, achieving both accuracy and real-time performance.
2Ease of operation
If analytical solution of forward kinematics is derived for parallel mechanism, then the system of high-order equations is obtained, but it is difficult to solve in real-time
Solution Approach 1:
The patent performs the complex analytical derivation and numerical solution in advance, storing results in a lookup table. During real-time operation, only simple table queries are needed, transforming the difficult real-time solving problem into an easy lookup operation.
Solution Approach 2:
The patent replaces the computational mechanism (numerical solving of high-order equations) with a data storage and retrieval mechanism (lookup table). This substitution transforms the problem from one requiring complex real-time computation to one requiring simple data access, achieving real-time control capability.
Data Source
AI summary
A robot control method, a legged robot using the same, and a computer-readable storage medium are provided. The method includes: obtaining a motion parameter of a driving mechanism of a target part of the robot; and obtaining an end pose of the target part by processing the motion parameter of the driving mechanism according to a preset forward kinematics solving model, where the forward kinematics solving model is a neural network model trained by a preset training sample set constructed according to a preset inverse kinematics function relationship. In this manner, a complex forward kinematics solving process can be transformed into a relatively simple inverse kinematics solving process and neural network model processing process, which reduces the computational complexity, shortens the computational time, thereby meeting the demand for real-time control of the robot.