Enabling automated driving for vehicles in the logistics industry
Viral Gosar defended his PhD thesis at the Department of Mechanical Engineering on March 26th.
The global use of automated vehicles, ranging from self-driving cars to mobile robots in warehouses and automated guided vehicles in ports, is increasing. In the European Union there鈥檚 great potential for these vehicles in, amongst other, the logistics industry. The tractor semi-trailer combinations that are being used in this industry get loaded and unloaded in distribution centers (DCs) with limited space to maneuver. Viral Gosar focuses in his PhD research on enabling automated driving for tractor semi-trailer combinations in DCs, emphasizing vehicle modeling, path-following control, and implementation in a scaled testing environment. The results of this research can be used as a foundation for future coordination of multiple vehicles within space-restricted environments.
The first step to enable automated driving for tractor semi-trailer combinations is to develop a vehicle model that can accurately and efficiently predict vehicle motion. This model serves as the base for controller design. To identify a suitable model, Viral Gosar conducted real-life tests with a full-scale vehicle. The collected data from these tests was then used to evaluate three different models withy varying levels of complexity. The model with the best balance between accuracy and efficiency has been used to develop a path-following controller that allows both forward driving and reverse docking. Its controller gains are based on vehicle dimensions, which makes it suitable for different vehicle sizes and configurations. The controller that Viral Gosar developed can heuristically determine whether the vehicle can dock with the required accuracy. If that is not possible, it initiates a forward movement before reversing, mimicking human driving behavior.
Path planning and coordinating multiple vehicles
In addition to individual vehicle control, algorithms for path planning and coordinating multiple vehicles have been created during this PhD research to analyze a DC environment with multiple tractor semi-trailer combinations. Viral Gosar evaluated these developments through simulations and tests in the TruckLab environment with scaled vehicles that have been calibrated using real-world data to closely match the driving characteristics of a full-scale tractor semi-trailer.
Remaining challenges
While this research offers tools to support automation in DCs, various challenges remain. This PhD research assumes, for example, that the location of the vehicle is precisely known. A perception system for detecting obstacles or mapping the environment has not been developed nor has dynamic path planning been introduced to avoid obstacles. Despite the remaining challenges it is expected that DCs will gradually employ tractor semi-trailers with increasing autonomy while reducing dependency on human drivers. This shift can help improve supply chain efficiency, lower logistics costs, and address labor shortages.
Title of PhD thesis: . Promotors: Prof. Henk Nijmeijer and Associate Prof. Igo Besselink. Co-promotor: Assistant Prof. Mohsen Alirezaei.