Robotic Grasping

Machine learning implementations using OpenAI Gym and PyBullet for robotic grasping.

Deep Q Learning (DQN)

• Discrete control problem is considered.
• Based on PyTorch tutorial example.
• kuka_rl

Proximal Policy Optimization (PPO)

• Continuous control problem is considered.
• kuka_rl_2

A2C Parallel Implementation

• Twice the number of available CPU threads is considered for the KUKA environment to run in parallel.
• kuka_rl_2_parallel

Dependencies

pip install pybullet
pip install gym