In this research, a neuro-fuzzy controller has been designed and implemented for a robotic leg system. The compound pendulum model uses all the required forces acting on the system, so it has been used to develop the basic model. The controller combines a neural network, which is used for the non-linear dynamic modelling, and a fuzzy logic system to model uncertainties with regards to the inputs. The simulations are done on the MATLAB/SIMULINK environment as it simplifies complex concepts through the intuitiveness of the program. A conventional PID controller has also been designed and the variable error of the output renders it unsuitable for modelling non-linear systems. The results show that the neuro-fuzzy system is the most suitable for the leg model, due to the immediate response time, accuracy, and stability. The controller successfully combines the strengths of these machine learning methods to accomplish high-level control. This reinforces the potential for further development of these control techniques for various robotic systems.