A four-legged robot foot end ground slip rate estimation method based on deep learning

By combining deep learning networks with GPS positioning and geological data, the problem of inaccurate slip ratio estimation in quadruped robots was solved, and the accuracy of terrain adaptability and friction conic coefficient adjustment was achieved, thus improving the robot's control performance in complex terrains.

CN115391680BActive Publication Date: 2026-06-05CHINA NORTH VEHICLE RES INST
View PDF 2 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN Β· China
Patent Type
Patents(China)
Current Assignee / Owner
CHINA NORTH VEHICLE RES INST
Filing Date
2022-08-16
Publication Date
2026-06-05

AI Technical Summary

Technical Problem

In the existing technology, the slip rate of the feet of quadruped robots is not accurately estimated on different terrains, which makes the adjustment of the friction conic coefficient unsuitable for complex terrains, and the traditional control strategy has poor robustness.

Method used

By combining deep learning networks with GPS positioning results and geological data, terrain semantic recognition and online estimation of slip rate are performed using robot sensor data. A classification and discrimination neural network is constructed to identify terrain attributes and obtain accurate slip rate.

Benefits of technology

It achieves highly reliable and sensitive terrain attribute judgment, improving the quadruped robot's adaptability to different terrains and the accuracy of friction conic coefficient adjustment.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN115391680B_ABST
    Figure CN115391680B_ABST
Patent Text Reader

Abstract

The application belongs to the technical field of automatic control, and particularly relates to a four-legged robot foot end ground slip rate estimation method based on deep learning, which adopts a deep neural network to realize judgment on a current terrain attribute category, further adopts a GPS global positioning result to obtain a geological data set under corresponding latitude and longitude, so as to obtain accurate terrain slip rates of the same terrain attribute in different regions; the application applies deep learning technology to four-legged robot foot end ground slip rate estimation, solves the problems of complex, multiple parameters and poor robustness of a traditional artificial landing judgment strategy, obtains a terrain attribute judgment neural network with high reliability and high sensitivity through long-term data collection and training, further combines a geological positioning data set to obtain accurate slip rates, and the method has good generalization and migration ability on different robot configurations and platforms.
Need to check novelty before this filing date? Find Prior Art