Sensor-based approach for manipulation tasks execution with a collaborative robot

Abstract

The quantity of domestic waste is constantly increasing, resulting in an overload of wastes. Recycling became imperative for ecologic and economical considerations in order to apply cir-cular economy rules. In general, humans are recruited in waste sorting plants for performing the dangerous and tedious sorting tasks. Automation is thus necessary. This report deals with the problem of waste sorting automation. In this work, we propose a robotised automated waste sorting system with the ability to detect human presence and co-operate with the human operator when needed. In order to accomplish the solution, We used computer vision based waste recognition and classificatio nusin gYOLOv8 ,huma npresence detection using RFID (Radio Frequency Identification )tag swhic hallow shuman-robo tcol-laboration and a robotic arm to pick and place waste objects. The system was built in ROS (Robot Operating System), starting the work with a virtual environment for validation before implementing the system on a real collaborative robotic arm. The system’ implementation led to good results concerning the waste detection part. Objects classes were identifie daccurately with acceptable precision in object positioning. The human detection was accurate within the range of RFID modules and robot movements were executed without any issues.