Humanoid Robot Nao: Developing Behaviours for Soccer Humanoid Robots

The topic of humanoid robot control has driven much research in areas such as robotics, artificial intelligence, computer vision, among others. Nowadays, humanoids are equipped with several degrees of freedom leading to very versatile movements but also to more complexity regarding the inherent problems. Robocup is a robotic soccer international competition that promotes the research in these areas, including humanoid leagues, both on simulated and real environments. The FCPortugal team has already a good record for wins in RoboCup. The team intends to study these problems due to its participation in these humanoid leagues. Therefore, this dissertation proposes the development of automated processes to optimize behaviours and generate sequences of behaviours for a given task regarding the environment. An optimization engine was extended and improved. Several kinds of kicks were developed and optimized using the Hill Climbing and Tabu Search methods. Also a planning approach was used to sequence different behaviours, providing a solution for avoiding obstacles and precisely control the position and orientation of the agent.