Efficient and Risk-Aware Framework for Autonomous Navigation in Resource-Constrained Configurations

Abstract

Path planning is a key challenge for autonomous vehicles, requiring solutions that balance safety and efficiency. This article proposes an autonomous road navigation system that does not rely on precise GPS, HD maps, or high-speed communication, making it particularly suitable for sparsely urbanized rural areas. The proposed method uses a tentacle-based path planning algorithm to compute the fastest possible trajectory while ensuring safety. A real-time traversability map, built and continuously updated from LiDAR (or alternative sensor) data, allows the robot to dynamically assess the risk of collision. The algorithm accounts for sensor perception limits, ensuring that any new obstacle appearing beyond the sensor range will not cause a collision. Simulation results are presented to evaluate and demonstrate our approach’s ability to simultaneously optimize speed while ensuring safety garentees.