Robot Learning Project

This project explores how a robot can learn to interact with objects using Reinforcement Learning. Specifically, we aim to train the Adeept 5-DOF Robotic Arm to open a box lid and/or grasp an object through trial-and-error learning.

---

Project Goals

• Build and control a 5-DOF robotic arm
• Use a camera to detect object states (e.g. lid angle, object position)
• Train a reinforcement learning agent to perform simple manipulation tasks

---

Getting Started

1. Clone the repository

git clone https://github.com/your-username/robot-learning-project.git
cd robot-learning-project

2. Set up the environment

We use uv for dependency management. To install uv go to https://docs.astral.sh/uv/getting-started/installation/

uv venv
uv pip install -e .

3. Run Q-learning

uv run -m rl.train