Machine Learning has been an area of humungous research recently, nowadays reinforcement learning is being widely used in solving gaming problems and in other domains too. Prosthetic is one such field where Reinforcement Learning Algorithms take intelligent decisions and are being use to imitate natural human movement. Amongst these prosthetic devices is prosthetic knee where the patient/amputee always long for a natural gait. Prosthetic Knee is a replacement of a natural leg which is lost due to multiple reasons such as the diabetes, peripheral vascular diseases to name a few. Number of lives which is affected by amputation is huge. The amputee is not only deprived of physical capabilities, but psychological setback. With the advent of sensor driven devices which are personalized as per the users and Artificial Intelligence being use to develop real world applications, it is believed that human life will be easy. In this paper, a survey has been done on the work done until now in developing a prosthetic knee joint and a review is done as to how Reinforcement Learning is impacting lives and being used in real time applications and out the best performing agent for prosthesis.