GNSS Position-Based Heading Estimation of Cut-to-Length Harvesters in Thinning Operations under GNSS-RTK-Denied Environments

Sammanfattning

Knowing the position of a harvester in the forest enables operators to utilize specific assistance systems, such as geofencing or path-following. The most effective solution for outdoor use is GNSS. However, harvesters operate under harsh conditions that negatively affect GNSS capability. One of the most prominent challenges for a forestry machine manufacturer is the use of GNSS for reliable heading estimation. To enable robust heading estimation for geofencing, specifically for the harvester head, and other future applications, a trajectory-based heading estimation method using harvester-specific data and GNSS position has been developed. In addition, an attempt is made to estimate different metrics for each method that can be used as trust values. Finally, a comparison is made with a visual odometry method, as this is considered a robust approach for bridging heading estimation in GNSS-denied environments.The results show that using GNSS position data for heading estimation comes with major drawbacks and does not provide a solid foundation, particularly when a reliable trust metric is required. In contrast, visual odometry demonstrates robust behavior with only small drift over time. Consequently, it is suggested to implement a more hardware-intensive solution utilizing visual-inertial odometry in a harvester.