Deep Reinforcement Learning Autonomous Warehouse AGV Navigation
Proximal Policy Optimization (PPO) agent trained in PyBullet to navigate dynamic factory obstacles without hand-crafted heuristics.
Project Overview
Trains an automated guided vehicle (AGV) in a continuous 3D physics environment. The agent takes 2D simulated LiDAR range findings as input and outputs continuous steering angles and throttle, learning collision-free shortest-path navigation in dense pedestrian warehouses.
Trains an automated guided vehicle (AGV) in a continuous 3D physics environment. The agent takes 2D simulated LiDAR range findings as input and outputs continuous steering angles and throttle, learning collision-free shortest-path navigation in dense pedestrian warehouses.
Proximal Policy Optimization (PPO) agent trained in PyBullet to navigate dynamic factory obstacles without hand-crafted heuristics.