Robot Vacuum Cleaner Low Obstacle Detection And Avoidance Based On 3D ToF

There is the general trend of smart homes.As a representative of service robots,robot vacuum cleaners are gradually entering ordinary households.At present,mainstream vacuum cleaner products are equipped with a single line lidar ranging system,which can locate position,mapping,path planning and executing in real time.They already have relatively mature capabilities of autonomous planning and cleaning.However,most of the robot vacuum cleaners currently on the market have issues with common low obstacles at home when they perform the cleaning process,such as slippers,wires,keys,etc.The obstacle avoidance effect is poor.It is easy to hit low obstacles or get stuck and unable to escape.Such a thing would require manual operation to get out of trouble,even the robot would be damaged.The reason is affected by the lidar of robot vacuum cleaners.To ensure the effect of mapping,lidar is installed on the top of the robot vacuum cleaner,therefore lidar cannot observe the low obstacles.So,the robot vacuum cleaner can only rely on the front collision sensor to contact the low obstacles before the rough perception.At the same time,because robot vacuum cleaners cannot predict the existence of low obstacles,it is difficult for manufacturers to effectively make obstacle avoidance strategies and algorithms.How to effectively detect low obstacles and avoid them when performing the cleaning job is one of the key research problems in the field of robot vacuum cleaners and has great potential application value.Aiming at the problem that the lidar sensors of current lidar navigation robot vacuum cleaners cannot effectively observe and recognize low obstacle,this thesis designed and implemented a robot vacuum cleaner low obstacle detection and avoidance prototype system based on 3D ToF(Time of Flight)technology,including hardware construction and software development.This system implements two main functions of low obstacle recognition and avoidance.The main work accomplished in this thesis are summarized as follows:(1)Implementation of low obstacle recognition algorithm and software.This thesis analyzed the principles and advantages of 3D ToF technology,which is commonly used in 3D scene perception.In this thesis,a low obstacle recognition algorithm and software module based on 3D ToF was proposed.The research and development work includes 3D ToF raw data acquisition,data preprocessing,point cloud generation,point cloud processing,and result output.(2)Implementation of low obstacle avoidance algorithm and software.Based on low obstacle recognition,a low obstacle avoidance algorithm and software module was proposed.This research and development work includes these parts:a multi-level costmap module which include low obstacle information and lidar sensor messages;a module which dynamically switches the robot mode between normal path follow mode and dynamic low obstacle avoidance mode;and a module of low obstacle avoidance implementation.(3)Based on TurtleBot 2 mobile robot platform and low-cost 3D ToF sensor,a robot vacuum cleaner prototype system including hardware and software was proposed.The prototype system is tested and analyzed in both simulation and real environment.Experiments show the system designed in this thesis can overcome the limitation that the laser cannot perceive low obstacles on such lidar navigation robot vacuum cleaners.And the result proves the effectiveness of robot vacuum cleaners in recognition and avoiding low obstacles when performing cleaning jobs.