Autonomous Excavator System with Real-World Deployment

—Excavators are widely used for material-handling applications in unstructured environments, including mining and construction sites. Workers operating excavators suffer from prolonged working hours and loads, which can result in injuries and fatalities. In this paper, we highlight our recent progress on developing autonomous excavator systems (AES) for material loading tasks. We present an architecture that combines perception, planning and control. We fuse multi-modal perception sensors, including LiDAR and cameras, with advanced image enhancement, material and texture classifica-tion, object detection, terrain traversability mapping, motion planning, and terrain navigation algorithms. AES has been successfully deployed in a real-world scenario, where two excavators automatically operate in recycling pipelines and handle hazardous industrial solid waste material. AES can achieve 24 hours of continuous operation for the scenario and has been used by the customer for more than 8 , 500 hours.