Design And Implementation Of Remote Interaction Control And Active Obstacle Avoidance System On Robots

With the development of artificial intelligence,the research of home robot is also deepened continuously,among which the robot's behavior of walking and obstacle avoidance based on multi-sensor information fusion and environment feature types has become a hot topic.However,traditional home robot has many problems,such as the poorly interaction of remote data,single data processing mode of obstacle avoidance,and insufficient use of environmental data.Therefore,in this paper,the construction of the robot system,the recognition of the environment type and the obstacle avoidance algorithm are studied.Aiming at the problem of poorly interactive of remote data,a remote control system based on video and heat map data is set up.According to the problem of insufficient use of environment type data,the environmental data is sensed by various types of sensors and the effective identification of environment type is completed.And an algorithm of robot avoiding obstacle based on environment perception and adaptive fuzzy neural network(FNN)is proposed.Firstly,this paper designs and builds a remote robot control system based on Web with HCR robot development board which was designed by DF-Robot company and raspberry Pi.We adds many types of sensors on HCR,such as ultrasonic sensors,infra-red sensors,crash sensors.Then the system can realize the function of flexible remote information exchange,remote control and so on.Additionally,the ECharts visualization tool is introduced into the system,so the information of obstacle distance can be displayed on the remote control terminal as thermal and radar chart.Secondly,a new environment type recognition method is proposed basing on based on environment information perception,We fully exploit all kinds of sensors to collect the environmental data,so that the robot can effectively identify the specific environmental types,such as wall-following state,Slope changing state,land subsidence state and risk-state.This method can also provide the data support for the emergency braking of the robot when it is in danger and the research of the following obstacle avoidance algorithm.Thirdly,basing on the former works,two obstacle avoidance algorithms of robot basing on fuzzy system(FUZZY)and adaptive fuzzy neural network(ANFIS)is studied and realized.According to the simulation of training data,both of these algorithms can effectively achieve the obstacle avoidance of the robot,and the obstacle avoidance algorithm based on ANFIS is better than FUZZY algorithm,because its control of robot walking velocity is more accurate.Finally,combining the environment recognition method and ANFIS algorithm,we propose a new algorithm based on the environment type perception and adaptive fuzzy neural network.We take EP-FNN algorithm as abbreviation of Robot obstacle avoidance based on environment perception and FNN.The obstacle avoidance strategy from FNN will be readjusted by the environment perception.After the software simulation of EP-FNN,it could be found that the robot walking velocity is more careful and safe.This algorithm effectively overcomes the shortcomings of single robotic obstacle avoidance type of obstruction or directly ignorance of the environmental impact.