Research on the Approaches and Opportunities of a Free Space Detection Function applied to Trucks

The primary objective of this thesis is to conduct research on Free Space Detection (FSD) functions and evaluate the opportunities and applicability of it for trucks, considering their unique operational and environmental challenges. The thesis focuses on adapting and evaluating existing FSD methods from passenger vehicles and robotics. The research started with a review of the state of the art in sensor technologies and existing algorithms to a detailed analysis of potential applications, requirements and integration for trucks. Starting with the state of the art, this thesis reviews various sensor technologies including LiDAR sensors, radar sensors, cameras and ultrasonic sensors. The review explains the functionality and highlights the strengths and weaknesses of each sensor, followed by a review of existing FSD, Occupancy Grid Mapping (OGM) and object detection methods, including approaches such as “NVRadar-Net”, “Dynamic Occupancy Grid Mapping with Recurrent Neural Networks”, and “YOLOv10”. After that some Advanced Driver Assistance Systems (ADAS) are described and then some sensor fusion techniques are introduced. The comparison of FSD with other ADAS shows that integrating FSD with functions such as Adaptive Cruise Control (ACC), Blind Spot Detection (BSD) or Lane Keeping Assist (LKA) can be highly beneficial and crucial for enhancing the overall safety and efficiency of trucks. It is also shown that FSD can be useful for various use cases, including lost cargo detection, parking space detection or in low-speed maneuvering scenarios. The thesis then outlines specific requirements for implementing FSD systems in trucks, considering factors such as perception, dealing with requirements for each sensor type, sensor placement, discussing the requirements for the placement of each sensor to work properly for FSD, system, ensuring accurate detection and response times, OGM, focusing on precise environmental mapping and truck requirements. The integration of FSD systems for trucks get explored, focusing on compatibility with other ADAS functions and necessary modifications to address certain limitations. In conclusion, it becomes clear early in the research that no publicly available FSD solutions exist today which are specifically designed for trucks. Current approaches, while effective for passenger cars, cannot be used for trucks without further modifications. Future research directions should involve adapting the current approaches for trucks, optimizing the machine learning models for largescale vehicles and investigating sensor fusion techniques specifically for sensor suites for trucks.