Eva Reitbauer and Christoph Schmied, Institute of Geodesy, Graz University of Technology, Austria; Michael Schedler, Institute of Logistics Engineering, Graz University of Technology, Austria

View Abstract Sign in for premium content


In the most common method of commercial composting, bio-waste is stacked into triangle-shaped windrows. As oxygen is needed for the composting process, the windrows need to be turned repeatedly by compost turners with spiked drums. The biogenic material has to be turned at low speeds and during the turning process, the drivers are exposed to water vapor, gases and high temperatures. Due to these unpleasant working conditions, operators of large composting plants often have difficulty in finding personnel. Therefore, autonomous compost turners would be a great benefit for operators of composting plants. A prerequisite for an autonomous compost turner is a robust and reliable positioning algorithm which is capable of providing continuous and precise position, velocity and attitude information of the compost turner in real-time. This paper presents a positioning module for tracked compost turners. It gives an overview on the system design and evaluates which sensors can be used for positioning of compost turners based on tests in the environment of a real composting site. Based on the preliminary sensor tests, a set of suitable navigation sensors is selected. The selected sensors include a GNSS dual-antenna array, a stereo camera, a MEMS IMU and encoders. The images recorded by the stereo camera are fused with a precise 3D map. Within this paper, a federated loosely coupled extended Kalman filter is developed to fuse the result of the stereo-image and model matching with GNSS, INS and odometer measurements. The primary innovation lies in the sensor selection and sensor fusion specifically tailored to compost turners. A novel approach, which fuses GNSS, INS, odometer, stereo images and a 3D map, is presented.