pollux-AMR

An autonomous countertop-cleaning robot with integrated reinforcement learning to prevent falling off cliffs and hitting obstacles.

Stack Architecture:

1. Python 3.10.7 for coding & development

2. Stable Baslines3 (SB3) (built on PyTorch) for the RL library

3. OpenAI Gymnasium (Gym) for simulation

4. ROS (Robot Operating System) Noetic for facilitating communication between the sensors/hardware, the RL libraries, and the simulation tools

5. Development Environment: Windows PC with WSL2 running Ubuntu 20.04

Hardware Components:

1. Robot Base: 2 wheels, 1 castor wheel

2. Sensors: 4 downward-pointing HC-SR04 ultrasonic sensors for cliff detection

   • (Optional) Stereo camera or two Raspberry Pi camera modules