Road accidents involving pedestrians remain a leading cause of mortality worldwide. As autonomous driving technology advances to address this issue, Deep Reinforcement Learning (DRL) has emerged as a promising approach for teaching vehicles to navigate safely, particularly in avoiding pedestrians. This paper presents a DRL algorithm for pedestrian avoidance in autonomous four-wheeled vehicles, simulated using Webots and Deepbots framework. This study employs stochastic and deterministic algorithms to train the vehicle in specific simulated scenarios. The study also aims to enhance pedestrian avoidance capabilities, advance autonomous vehicle safety, and address a challenge in fully autonomous vehicles capable of drivers' and pedestrians' safety. The study also aims to provide a fundamental approach for further improvement on safer autonomous driving DRL algorithms in Webots simulators in the future.
Pedestrian Avoidance Simulation by Deep Reinforcement Learning Using Webots
18.02.2025
1005459 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
Pedestrian Collision Avoidance Using Deep Reinforcement Learning
Springer Verlag | 2022
|