Design Of Intelligent Garbage Sorting And Transportation Robot Based On ROS

With the continuous improvement of people’s living standard,it is unavoidable to increase the discharge of household garbage.Garbage classification is an effective way to solve this problem,but the supervision and guidance of garbage classification and the cleaning and handling of garbage require a lot of manpower.Therefore,we design and implement an intelligent garbage sorting and transporting robot based on Robot Operating System(ROS)by applying advanced technologies such as depth learning and autonomous navigation.It can complete a series of functions such as picking up,classifying,transporting and placing garbage,and it is believed that it will have a broad development prospect.The main research contents of this paper are as follows:(1)The design,modeling and Simulation of the robot system are completed.The actual functional requirements of the robot are analyzed,and the hardware structure and software framework of the robot system are designed.The kinematics and dynamics of the mobile chassis and the manipulator are analyzed,and a Gazebo simulation model of the robot is built for the simulation test.(2)The function of garbage identification,location and collection is designed and implemented.The relevant theories of binocular camera and ssd＿inception＿v2 model are analyzed.The garbage image data set of this paper is built and the model training is carried out by using transfer learning algorithm.A scheme of garbage space location is designed and a filtering algorithm is proposed to optimize it,which improves the efficiency of target location information.A strategy of garbage collection and classified placing is designed and optimized.A shz＿target message type is proposed to ensure the correct garbage classification and a trigger mechanism is proposed to reduce the consumption of computer resources by garbage identification.(3)A complete coverage path planning and navigation obstacle avoidance function is designed and implemented.The principle of Gmapping mapping and the comparison and selection of two complete coverage algorithms are discussed.Adaptive Monte Carlo Localization(AMCL)positioning,navigation framework,and Time Elastic Band(TEB)local path planning principles are studied.A complete coverage navigation algorithm which combines navigation stack and PID control method based on circle detection area is presented to improve the complete coverage effect of the robot.On this basis,an automatic obstacle avoidance method combining local cost map and TEB algorithm is presented to avoid obstacles of different sizes.The integration of various functions is Completed.The human machine interaction interface is designed to control and observe the status of the robot conveniently.Experiments are carried out in simulation and real environment.The experimental results show that the positioning accuracy of garbage in X,y and Z directions are 4.73 mm,3.85 mm and 2.96 mm respectively,and the success rates of target garbage recognition and sorting are 90.56% and85.00% respectively.The spatial positioning efficiency using the filtering algorithm in this paper is 90.07%,which is 10.45% higher than that before optimization.The coverage and repetition rate of full coverage path planning in the actual environment are 80.27% and 6.39%respectively.The experimental results show that the intelligent garbage sorting and transporting robot based on ROS designed and implemented in this paper works well and meets the expected requirements,which lays a foundation for subsequent related research.