Aaronaferns/MBRL-DeepMPC

Efficient Model-Based Deep Reinforcement Learning with Predictive Control: Developed a Model-Based RL algorithm using MPC, achieving convergence in 200 episodes (best case) and 1000 episodes on average, outperforming SAC/DQN (10,000+ episodes). Enhanced sample efficiency by 80-90% using learned dynamics and CEM for trajectory optimization.

20
/ 100
Experimental

This project helps robotics engineers and control systems designers efficiently train intelligent agents for continuous control tasks, such as balancing a pole on a cart. It takes in real-time observations from a robotic system and outputs optimized action sequences to achieve desired behaviors. The primary users are researchers and practitioners working with robotic systems that require precise control.

No commits in the last 6 months.

Use this if you need to train a robotic system to perform continuous control tasks with high sample efficiency, especially in environments where rewards are dense and immediate.

Not ideal if your robotic control problem involves sparse rewards or if you are working with very high-dimensional systems where computational cost is a major concern.

robotics-control reinforcement-learning motion-planning automation predictive-control
No License Stale 6m No Package No Dependents
Maintenance 0 / 25
Adoption 5 / 25
Maturity 8 / 25
Community 7 / 25

How are scores calculated?

Stars

10

Forks

1

Language

Python

License

Last pushed

Dec 31, 2024

Commits (30d)

0

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/ml-frameworks/Aaronaferns/MBRL-DeepMPC"

Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.