Navigation In Urban Environments Amongst Pedestrians Using Multi-Objective Deep Reinforcement Learning

Urban autonomous driving in the presence of pedestrians as vulnerable road\nusers is still a challenging and less examined research problem. This work\nformulates navigation in urban environments as a multi objective reinforcement\nlearning problem. A deep learning variant of thresholded lexicographic\nQ-learning is presented for autonomous navigation amongst pedestrians. The\nmulti objective DQN agent is trained on a custom urban environment developed in\nCARLA simulator. The proposed method is evaluated by comparing it with a single\nobjective DQN variant on known and unknown environments. Evaluation results\nshow that the proposed method outperforms the single objective DQN variant with\nrespect to all aspects.\n