Lidar-based Obstacle Detection and Recognition for Autonomous Agricultural Vehicles

Abstract

This thesis would not have been possible without support from a great number of people. First, I would like to thank my main supervisor Rasmus Nyholm Jørgensen for offering me the PhD position and for always seeing (financial) possibilities instead of limitations. You have always defended our interests and helped out during planning and execution of difficult field trials. Thanks to my co-supervisor Henrik Karstoft. You have always been available for technical guidance, discussions, and constructive feedback, and you have been a great help in soliciting my next position. I look forward to our continuous collaboration. Also a huge thanks to my previous co-supervisor Henrik Pedersen for the guidance and for facilitating previous, current, and future positions in the field of computer vision. I would further like to thank my supervisor at the Australian Centre for Field Robotics, James Underwood, for his great guidance, technical know-how, and willingness to brainstorm and discuss new ideas even years after our collaboration. I have enjoyed every minute of the research stay, and I sincerely hope to cross paths again in the future. Thanks to all my great colleagues, Anders, Martin, Mads, Søren, Simon, and Rene for continuous feedback and technical discussions. Thanks to Morten Larsen and Morten Laursen for your help in choosing, interfacing, and debugging sensors and hardware for the perception platform. A special thanks to my project partner and friend Peter Christiansen for great collaboration and helpful discussions. Last, but certainly not least, a special thanks to my loving wife, Astrid. This work would not have been possible without your endless support, patience, and understanding through the entire study. Thank you for joining me on the research stay abroad, and thank you for making it all run smoothly at home when I have been unpredictable and buried in work.

Sensing Technology Survey for Obstacle Detection in Vegetation

“…Point cloud classification deals with the issue of discriminating 3D point structures based on their shapes and neighborhoods for various applications, such as vehicle identification and tracking [23], pedestrian-vehicle near-miss detection [24], and background filtering [25]. Kragh [26] proposed two methods for point classification of LiDAR-acquired 3D point clouds, which address sparsity and local point neighborhoods and were used for consistent feature extraction across entire point clouds. One method, based on a traditional processing pipeline, outperformed a generic 3D feature descriptor designed for dense point clouds.…”

“…Together, the two methods showed that sparsity in LiDAR-acquired point clouds can be addressed intelligently by utilizing the known sample patterns. A combination of multiple representations may therefore accumulate the benefits and potentially provide increased accuracy and robustness [26]. To effectively use LiDAR for sensing obstacles in vegetation, LiDAR can be used in combination with stereo cameras to analyze the cloud data points with 2D images.…”

“…It addresses the question of how LiDAR technology can work with other sensing modalities in agricultural environments. Kragh [26] proposed and evaluated methods for sensor fusion between LiDAR and other sensors, incorporating spatial, temporal, and multimodal relationships to increase detection accuracy and thus enhance the safety of self-driving vehicles. The method consists of a self-supervised classification system using LiDAR to continuously supervise a visual classifier of traversability.…”

“…Comparatively, stereo cameras can detect and sense obstacle depth, as well as identify camouflaged animals and protruding objects. A key ability of thermal cameras is that the sensing data are not affected by camouflaged animals and the illumination effect while the performance capabilities are affected by the ambient temperature [26]. Another drawback of thermal cameras is the low resolution and loss of range data when the camera is in motion, as well as when there is texture difference.…”

Multimodal obstacle detection in unstructured environments with conditional random fields

“…This ensures high resolution at short distance and prevents noisy features at far distance. We use the method from Kragh, Jørgensen and Pedersen () which in Kragh () has shown to outperform the generalized 3D feature descriptor fast point feature histograms (FPFH) (Rusu, Blodow, & Beetz, ) for sparse, lidar‐acquired point clouds. The method scales the neighborhood size linearly with the sensor distance.…”