Research On Robot Autonomous Avoidance Control System In Human-robot Coexistence Environment

In recent years,with the continuous advancement of technologies such as artificial intelligence,more and more service robots are integrated into our lives and provided services for human,human-robot coexistence has become a future trend.In the home environment where people and robots coexistence,on the one hand robots should provide services at acceptable distances of human psychologically;on the other hand,for the safety of human,robots must be avoided human when necessary to avoid possible hurt.Robot autonomous avoidance becomes a prerequisite for their widespread use in the home.This paper studies the robotic autonomous avoidance control system in the human settlement environment.The study uses global vision to realize real-time detection and localization of human body in indoor environments,and realizes robot's autonomous avoidance to people.This paper constructed the robot's autonomous avoidance control system,completed the research on the overall composition of the system,the data transmission and the robot's avoidance strategy considering human-robot relationship.The robotic autonomous avoidance control system is mainly composed of visual inspection system,an indoor omnidirectional mobile robot,server and wireless network.In the visual inspection system realized detection and localization of the human and the robot,and based on the human-robot distance and the avoidance strategy generate robot avoidance tasks,then transmit it to the server via the wireless network;The server transmits the task to the robot through the wireless network,and the robot executes the command to complete the avoidance movement.This paper analyzes the common human detection algorithms in human settlements based on vision and then proposes a human detection scheme based on human body depth information and joint point matching.The vision system uses the depth image to detect the human body.First performs image preprocessing,obtains the region of interest,uses the bone point model to match,and take the position of the spine joint as the position of the human body in threedimensional space.Then through coordinate system transformation realized image mapping of human body position.The kernel correlation filtering algorithm based on HOG feature was used to track human motion,and the experimental verification was carried out.Secondly,an indoor omnidirectional mobile robot and a method based on it's color mark positioning are constructed.Achieved motion modeling,control system analysis and motion control of a mobile robot platform based on the Mecanum wheels.Through the color space transformation to reduce the influence of illumination and other factors on the recognition in visual positioning.Then,the histogram equalization,color segmentation and edge detection are used to detect the color mark of the robot,and combined the distance information of the depth image to determine the robot location.Finally,this paper verifies the feasibility of the system through experiments.The main verification experiments are as follows: 1.The experiments of human body detection,research the detection of the human body in the case of partial occlusion,different side conditions,and compares with other algorithms to verify the real-time and accuracy of human detection;2.The experiment verification the robot detection and positioning by visual inspection system and analysis the error;3.Robot avoidance experiment.In the human-robot coexistence environment,the position of the human and the robot is detected by the visual inspection system.When the distance is less than the safe distance,perform the avoidance movement,verified the feasibility of the system.