Research On Human Upper Limb Motion Trajectory Prediction Algorithm In Human-robot Cooperation

With the continuous development of robotics technology,the application scenarios of robots are no longer limited to traditional industrial fields.More and more collaborative robots are being used to complete designated tasks together with humans in the same-shared space.These tasks are usually tasks that traditional robots cannot do,have high accuracy,and require high cooperation with humans.The scenario where humans and robots work together in a shared space is called human-machine collaboration.The premise for smooth human-machine cooperation is to ensure human safety.Robots need to adjust their paths in a timely manner according to the complex and ever-changing working environment to avoid collisions with humans.The human upper limb,as the main joint part of the human body for interaction with robots,can well reflect human operational intentions.Therefore,it is necessary to predict the motion trajectory of the human upper limb in human-robot cooperation.Understanding each other’s movement intentions between humans and robots can enable robots to improve work efficiency while ensuring human safety effectively.The thesis conducts research on human upper limb motion trajectory prediction algorithms in human-machine collaboration.The research content is as follows:(1)According to the kinematics structure characteristics of human upper limbs,analyze the relationship between joints,use Kinect vision sensor to obtain human motion information and feature extraction,and preprocess the data to build a human-robot collaboration dataset.(2)A human upper limb trajectory prediction framework based on multi algorithm fusion is proposed to address the issue of low accuracy of a single prediction algorithm.Different prediction algorithms are combined through fusion strategies.The framework is divided into online and offline stages,where the offline stage directly learns the parameters required for the prediction model from task data;In the online stage,based on partial trajectory classification,polynomial fitting algorithm,and typical trajectory prediction,the trajectory within a given time node is predicted.This algorithm not only enables understanding of human intentions,but also improves the accuracy of long-term prediction.(3)A robust Gaussian mixture regression prediction model is proposed to address the uncertainty of human motion.Firstly,an improved Gaussian mixture model is used to cluster human upper limb motion trajectories,and then combined with Gaussian mixture regression to obtain statistical values for future trajectories.The advantage of this algorithm is that the predicted result is not only the predicted value of the position,but also the probability distribution of all possible future motion trajectories of the upper limb,providing an estimation of the uncertainty of human motion.(4)In order to verify the effectiveness of the prediction algorithm proposed in this article,a human-machine collaboration experimental platform was established in the medical laboratory to record the upper limb motion trajectory of humans in pathological reagent detection tasks and create a human-machine collaboration dataset.The prediction performance of the proposed algorithm was verified through simulation experiments.Analysis of the experimental results showed that the prediction algorithm for human upper limb motion trajectory proposed in this article is feasible and effective,Ensure human safety and improve work efficiency in human-machine collaboration scenarios.