Aim : To implement a Genetic Algorithm for Policy Search For Solving Open AI ‘CartPole-v1’ Environment. Open AI ‘CartPole-v1’ Environment consists of a pole balanced on a cart which moves on a frictionless track. The system is controlled by applying a force of +1 and -1 on the cart. A reward of +1 is awarded for every time-step the pole remains upright. Episode ends when the pole is more than 15 degrees from the vertical or the cart is 2.4 units from the center. The goal was to collect a average reward of 300 over 100 population. This is considered the population from which the best strategy is extracted.

Tech Stack Used :

1. ) Open AI Gym : https://gym.openai.com/ For simulating the cartpole environment in python
2. ) DEAP : https://github.com/deap/deap For implementing Genetic Algorithm for policy search
3. )PyTorch : https://pytorch.org/ For implementing neural networks in python

#ai