This Project investigated the ability of a Reinforcement Learning path planning algorithm to minimise flight costs in an urban environment using Orographic Soaring.
The RL_Path_Planner.py file contains the code for the design of the Grid Map, the Markov Decision Process Environment Class and the implementation of PPO with an MLP policy to complete the design of the path planner. It also includes the code used to train, evaluate and test the algorithm with render capability.
The TestGridLayouts.py file contains the code for the different obstacle environments used in the control and variable wind-field experiments.