Recognition method and system for motion behavior of legged robot based on multi-modal perception

Abstract

The present invention discloses a method and system for recognizing the movement behavior of a legged robot based on multi-modal perception, wherein the method includes: based on various types of sensors, collecting information obtained by the legged robot repeatedly walking a preset distance on different types of ground The generated multi-modal sensing data information; the multi-modal sensing data information is divided into data sets according to different types of ground to obtain multi-modal sensing data sets of different ground types; multi-modal sensing data sets based on different ground types The sense data set trains the hidden Markov model of the viscous hierarchical Dirichlet process to obtain the hidden Markov model of the optimal viscous hierarchical Dirichlet process; input the information of the samples to be identified into the hidden Markov model of the optimal viscous hierarchical Dirichlet process The husband model is calculated, and the sum of the logarithmic likelihood function values ​​is calculated; based on the calculation results, the movement behavior recognition of the legged robot is carried out. In the implementation of the present invention, the reliability and accuracy of motion behavior recognition are improved.

Description

technical field [0001] The present invention relates to the technical field of robot motion behavior recognition, in particular to a multi-modal perception-based method and system for legged robot motion behavior recognition. Background technique [0002] Footed robots have good environmental adaptability, wide range of motion, and strong load capacity, and can perform tasks such as rugged mountain transportation, dangerous disaster rescue, and military reconnaissance. Therefore, being able to recognize in real time the motion behavior and motion characteristics of the legged robot walking on different types of ground (such as sand, cement, grass, wood, tiles, etc.) will be able to directly adjust the control and gait of the robot, and improve the environment. Adaptability and robustness of the system. Therefore, it is the key technology for the frontier research of legged robots to carry out motion behavior recognition of legged robots based on joint encoders, IMUs, joint ...

AI Technical Summary

Problems solved by technology

However, as the complexity of the environment and the diversity of tasks increase, there are mainly two problems: (1) only considering the single-modal sensing information, it is impossible to accurately realize the perception of the environment and the state estimation of the robot system, reducing the It improves the fault tolerance of motion behavior recognition; at the same time, (2) the use of parametric HMM modeling method will have the problems of uncertain number of recessive states and rapid transition of recessive states, which cannot be learned from complex sensor data. Actual latent patterns, reducing the accuracy of motor behavior recognition

Images