continuous-control

Deep Deterministic Policy Gradients (DDPG) for controlling Robotic hand to grasp ball

Simple implementation of SAC with PyTorch.

Example FRWR (PDDM) implementation with ReLAx

EEL6938-Spring2025: Trains RL agents (SAC, PPO, TD3, DDPG) to follow a speed profile using stable-baselines3; compares algorithms and hyperparameters via MAE, MSE, RMSE metrics.

An implementation of DDPG algorithm on OpenAI Gym Pendulum-v0 environment.

Example CEM implementation with ReLAx

Pytorch implementation of twin delayed deep deterministic policy gradients (TD3)

PPO implemented from scratch in PyTorch for continuous control on LunarLanderContinuous-v3 (Gymnasium).

Example PPO implementation with ReLAx

Example TRPO implementation with ReLAx

Udacity deep reinforcement learning continuous control project

3D autonomous drone gate racing with Soft Actor-Critic, curriculum learning and multi-drone transfer in PyTorch.

Model-Based RL Multi-Tasking with ReLAx

Example Random Shooting implementation with ReLAx

PyTorch implementation of SAC (Soft Actor-Critic)

Hybrid robotic table-tennis controller combining DDPG reinforcement learning with physics-based positioning in a PyBullet simulation.

RL-Odyssey is a research framework for continuous control that implements state-of-the-art RL algorithms (SAC, TD3, PPO, etc.) with clean experiment scripts and interactive notebooks.

Benchmarking PPO, DDPG, and SAC for vision-based continuous control in Gymnasium CarRacing, with multi-seed evaluation and track generalization.

Using RL to control a double-jointed robot to reach target locations