Research On Non-Contact Human-Robot Interaction Based On Sensor Ring Array

Since the concept of German industry 4.0 was present,the global manufacturing industry has been facing new upgrading,resulting in the demand for intelligent,agile,flexible,scalable and modular manufacturing system.The human-robot cooperative production mode applied to complex tasks has become a new development trend in industry.Compared with traditional industrial robots,cooperative robots share workspace with operators and independently perform operation tasks in an open environment,which puts forward higher requirements for the safety of human-robot interaction.How to ensure the safety of human-robot interaction is a key issue for the application of cooperative robots.Giving various sensing abilities to cooperative robots is an important means to realize robot intelligence and security.Reducing the cost of sensing system will also contribute to the promotion and application of cooperative robots in small and mediumsized enterprises.The thesis compares the existing solutions to realize safe human-robot interaction at home and abroad,integrates its performance,advantages and disadvantages,and then designs a non-contact robot body perception system based on proximity perception from the perspective of sensing space coverage and system cost,and develops self-occlusion detection,dynamic obstacle avoidance and drag teaching algorithms based on UR robot as the application platform.Firstly,an inclined sensor ring array was designed based on the minimum safety monitored distance,which is arranged on the robot links in pairs to realize the safety perception of the robot.The relationship between the sensor arrangement parameters and the safety sensing space was analyzed,and the optimal sensor configuration parameters were optimized based on the maximum coverage of the sensing space.Experiments showed that the sensor ring array can meet the requirements of safety monitored distance.Secondly,the forward and inverse kinematics model of UR10 robot was analyzed,and the cooperative positioning model of measured object based on sensor ring array was established.Aiming at the self-occlusion problem,a self-occlusion detection algorithm based on robot geometry was proposed to evaluate and judge whether the object detected by robot sensing system is an obstacle or robot body in real time.Experiments verified the feasibility and effectiveness of the self-occlusion detection algorithm.Thirdly,using the combination of artificial potential field method and danger field method,the gravitational field acting on the end of the robot was established at the target position,and the repulsive field acting on the connecting rod of the robot was established at the obstacle position,so as to evaluate the safety of the robot in real time and carry out dynamic obstacle avoidance planning.The joint simulation of MATLAB and Coppelia Sim showed that the dynamic obstacle avoidance algorithm can effectively avoid obstacles and track dynamic targets.Finally,a non-contact drag teaching method was proposed based on the sensor ring array.According to the measurement results of the sensors,the teaching motion is decomposed into a combination of repulsive motion,radial motion and axial motion for teaching and tracking step by step.The radial and axial teaching and tracking processes were realized by bang-bang control and PID control respectively.Simulation and experiments showed that the decomposed motion of the drag teaching method has coupling,and the motion compensated teaching hand can make the robot complete the drag teaching task smoothly.