LAUSR

For a sustainable operation of multiple Self-Guided Vehicles (SGVs) in a dynamic manufacturing environment, it is essential to guarantee collision-free and efficient navigation to the autonomous mobile platforms and safety to the surrounding subjects. To prevent from navigation failures, an SGV must avoid conflicts that constrain itself to abruptly brake or stop to avoid collisions. These inefficient conflicts result from unexpected changes in the configuration space or due to nearby unforeseen obstacle. In this paper, a navigation approach is proposed to adapt the global trajectory in order to reduce conflict occurrence while limiting energy consumption of the mobile platform. To generate such trajectory, first the collision risks are characterized using an objective risk perception parameter, the Time-To-Collision TTC, that rely on the kinematics of the egoSGV and the neighboring obstacles. Next, weighted Kernel Density Estimation (wKDE) defines the spatial distribution of conflict severity in configuration space. The defined zones are incorporated as a conflict layer in the global map. Then, a global trajectory planner algorithm is used to weigh between the length cost and conflict cost. Finally, to test the proposed solution, a simulation is performed in a factory-like environment, then an experiment is conducted with a real SGV. In comparison with the state-of-the-art geometrical path planning method, the results show that the proposed approach reduces navigation failures by up to 52%, while reducing the trajectory execution time by around up to 10 %. Also, the smoothness of the executed motion allowed to reduce energy consumption by over 12%.