rl4j - Deep Reinforcement Learning for the JVM (Deep-Q, A3C)

•    Java

---

Deep Reinforcement Learning for the JVM

reinforcement-learning deeplearning4j doom cartpole a3c dqn gym-java-client artificial-intelligence

essence - AutoDiff DAG constructor, built on numpy and Cython

•    Python

---

A directed acyclic computational graph builder, built from scratch on numpy and C, with auto-differentiation supported. This was not just another deep learning library, its clean code base was supposed to be read. Great for any one who want to learn about Backprop design in deep learning libraries.

machine-learning dropout lstm mnist lenet neural-turing-machines question-answering computational-graphs auto-differentiation convolutional-neural-networks convolutional-networks recurrent-neural-networks lstm-model deep-learning deep-q-network reinforcement-learning cartpole

ilqr - Iterative Linear Quadratic Regulator with auto-differentiatiable dynamics models

•    Python

---

This is an implementation of the Iterative Linear Quadratic Regulator (iLQR) for non-linear trajectory optimization based on Yuval Tassa's paper. It is compatible with both Python 2 and 3 and has built-in support for auto-differentiating both the dynamics model and the cost function using Theano.

theano cartpole mpc control-systems trajectory-optimization optimal-control ddp dynamics-models auto-differentiation pendulum trajectory-tracking differential-dynamic-programming model-predictive-control non-linear-optimization model-predictive-controller ilqg ilqr mpc-control

pddp - WIP implementation of Probabilistic Differential Dynamic Programming in PyTorch

•    Jupyter

---

Probabilistic Differential Dynamic Programming (PDDP) is a data-driven, probabilistic trajectory optimization framework for systems with unknown dynamics. This is an implementation of Yunpeng Pan and Evangelos A. Theodorou's paper in PyTorch, [1]. This is a work in progress and does not work/converge as is yet.

robotics deep-reinforcement-learning pytorch artificial-intelligence uncertainty cartpole mpc probabilistic-programming trajectory-optimization ddp gaussian-processes dynamics-models pendulum pddp differential-dynamic-programming bayesian-neural-networks model-predictive-control ilqg ilqr double-pendulum

safe-control-gym - PyBullet CartPole and Quadrotor environments—with CasADi symbolic a priori dynamics—for learning-based control and reinforcement learning

•    Python

---

Physics-based CartPole and Quadrotor Gym environments (using PyBullet) with symbolic a priori dynamics (using CasADi) for learning-based control, and model-free and model-based reinforcement learning (RL). These environments include (and evaluate) symbolic safety constraints and implement input, parameter, and dynamics disturbances to test the robustness and generalizability of control approaches.

control reinforcement-learning quadcopter robotics symbolic gym cartpole safety quadrotor robustness pybullet casadi