Learning control of robots

In the process of human learning, the brain which acts as a controller receive sensory signals from other parts of the body and undergo processing to generate information which will then be stored. Based on previous compilation of information present, the brain will map similar experiences and trigger responses causing the person to react accordingly to the situation. Using similar principle, the neural network controller processes past results to produce desired responses via its learning function. Being a dynamic close loop system, the learning controller enables the robot to adapt to unknown situations by regulating its output based on the error present. In the field of region tracking, this concept can be observable as the tool of the robotic arm gradually moves towards and into the desired boundary space, reducing the error between the actual and the anticipated position.This thesis provide a comprehensive study on the neural network in the robotic arm to form a sensory-to-motor output transformation and to improve the accuracy of region reaching control.Bachelor of Engineerin