Pechckin / MountainCar

Solution
5 stars 3 forks source link

Mountain Car from gym Open AI

Mountain Car ploblem - https://github.com/openai/gym/wiki/MountainCar-v0

Mountain Car Continuous ploblem - https://github.com/openai/gym/wiki/MountainCarContinuous-v0

Solved with Q-learning algorithm with e-greedy policy and discretization. Used modified reward acording to potentials - https://people.eecs.berkeley.edu/~pabbeel/cs287-fa09/readings/NgHaradaRussell-shaping-ICML1999.pdf .

Also solved with DDPG, DDPG links: https://arxiv.org/pdf/1509.02971.pdf, https://towardsdatascience.com/deep-deterministic-policy-gradients-explained-2d94655a9b7b