Reinforcement Learning (RL) is a form of machine learning technique that investigates how a smart agent should act in a particular context (i.e. environment) to be able to achieve the greatest cumulative reward. It is a field that finds answers for sequential decision-making in stochastic situations. The aim of robust RL algorithms is to develop a robust best policy that takes into account model (i.e. environment) uncertainty to reduce the best policy’s sensitivity in noisy contexts. In this research, we explore a fundamental challenge for learning RL agents for real-life applications, namely the situation where the context changes. Such changes require that the agent continuously monitor environmental features and modify the learned policies for efficient functioning. We implement and examine two different reinforcement learning algorithms, Q-Learning and SARSA, in the Lunar Lander game in Open AI Gym package. We test the robustness of Q-learning and SARSA algorithms by adding different types of noise to their environments. The maximum average reward for the last 100 successive episodes achieved by the Q-learning agent in this paper is 177. We perform a comparative examination of the two algorithms to know which algorithm operates better and how they deal with different types of noise. My code is available at https://github.com/aghorab/Ghorab_Lunarlander