Autonomous Driving on Skid Tracks for Forestry Machines

Since labor costs make up a significant portion of the total cost of ownership and due to the severe labor shortage, research and development have increasingly focused on automating mobile machines. Due to its technical and sociological aspects, the Forestry offers high potential for using (semi-) autonomous machines. A significant proportion of forestry work consists of recurring processes. This paper introduces a framework for autonomous driving on skid tracks with forestry machines. Forwarders navigate skid tracks to the felled trees, collect them, and return to the forest roads, where the sorted piles are stored. During this process, the driver primarily focuses on the loading process, with driving being a secondary task. Automating the driving process reduces the driver's workload and allows them to concentrate on the more critical tasks. The proposed system comprises four submodules: localization, object detection, path planning, and driving. In forestry environments, GNSS signal reception is limited due to the treetops, and the system utilizes an adapted feature SLAM method to determine the vehicle's relative position. ... mehrThe object detection module covers the surrounding environment, detecting obstacles, such as stems and stumps, crossing persons, and the path of the skid track. Path planning uses the output of object detection to find a suitable path for the vehicle, while the driving module controls the actual steering and velocity. The presented system is implemented on an HSM Forwarder 208f, and its functionality is shown in a proof of concept on a skid track. The results prove that the autonomous system can relieve the driver of the driving task. The performance of the autonomous driving system is similar to that of a human driver, and the modules can be executed on currently available embedded hardware in real time.