| Abstract |
This project addresses the limitations of traditional wheeled and tracked search-and-rescue (SAR) robots, which often struggle to navigate extremely rocky or uneven terrain. To overcome these challenges, we propose the design of a quadrupedal spider-inspired robot optimized for agile and efficient locomotion in complex environments. The robot is trained using reinforcement learning, specifically Proximal Policy Optimization (PPO), to maximize its speed, efficiency, and balance when traversing challenging terrain. Due to resource constraints, the robot operates exclusively within a physics simulator rather than in a physical environment. |
