Project Overview

This REU project explores how robots can learn intelligent behaviors in physics-based simulation environments using imitation learning and reinforcement learning. Students will use the Webots platform to develop realistic robotic simulations that capture physical interactions such as dynamics, collisions, and contact forces.

The project integrates depth camera sensing to enable robots to perceive their environment in 3D. Using simulated depth data, students will develop perception pipelines for tasks such as object recognition, obstacle avoidance, and spatial mapping. These perception capabilities will serve as inputs for learning-based control strategies.

Through this experience, participants will gain hands-on skills in robotics simulation, depth sensing, machine learning, and programming (Python), while developing an understanding of how perception and learning interact in intelligent robotic systems. The project is relevant to applications in autonomous systems, smart manufacturing, and adaptive human-robot collaboration.