Deep Deterministic Policy Gradient and Computer Vision for Autonomous Wheelchair Navigation in Human-Shared Environments

We present an innovative approach to autonomous navigation for motorized wheelchairs in shared indoor spaces with human presence, combining Deep Reinforcement Learning and Computer Vision. The primary aim of this study is to enhance the well-being of individuals with disabilities who rely on such assistance for their mobility needs. Our results demonstrate that the Deep Deterministic Policy Gradient (DDPG) algorithm combined with computer vision techniques outperforms the Deep Q-Network (DQN) in both efficiency and stability. Throughout all analyzed stages, DDPG achieved higher average success rates: 98% (Stage 01), 89% (Stage 02), 86% (Stage 03), and 86% (Stage 04), showcasing generalization capabilities and consistently superior performance in diverse settings. In contrast, DQN struggled to avoid collisions, resulting in significantly lower average success rates: 3% (Stage 02), 14% (Stage 03), and 29% (Stage 04). These findings highlight the promising potential of our proposed solution and contribute to the advancement of research in this field.