Due to the similarity between human beings and humanoid robots, this kind of robot can serve in diverse applications instead of human such as servant for elderly people, working in hazardous environments and even as a robotic soldier. Therefore, in recent years, a great number of researchers in all over the world have been attracted to this field and many studies have been focused on this subject. Although, one of the priorities of this robot in comparison to others is its capability in crossing rough terrains, the challenging problem of stable walking makes the progress in the field slow, in a way with a small bump or pit in the ground profile might cause the robot to fall. However, for biped robots to exist around, they must be able to roboustly cross rough terrains. The aim of this thesis is to improve the movement of biped robots on rough terrains. To this aim a seven link planar biped robot equipped with four force sensors under the feet has been considered. These four sensors are placed beneath the tow and heel of each foot and their data are used to detect the ground profile during robot movements in rough terrain environments. Trajectory planning of biped has been done assuming the walking is on a flat surface. In this research a position/force controller is proposed which modify the initial trajectory based on the errors in tracking of desired forces. This controller consists of three main parts. The first part of controller is a closed loop position controller to track the desired trajectory, the second one is an open loop force controller to compensate for applied force on the feet, and the third part is a closed loop adjustable force controller. In case of error in the applied force on the feet, the controller has the property to automatically modify the base trajectory of the robot to compensate the error in the force tracking. This modification can improve the adaptability of robot to different ground profiles; i.e. an 80 percent improvement in step down situation and a 16 percent improvement in step up. In order to further improve the movement of the robot in rough terrain another controller is also proposed. This controller has the ability to independently control the trajectory of the hip joint and absolute angle of the trunk, while possessing other properties of the previous position/force controller. Numerical simulations shows that with this new controller the performance improves by 340 percent in step down and 133 percent in step up motion. Due to the fact that all of the proposed controllers are model based controllers, equations of motion of robot are required. We used MATLAB as a symbolic tool to derive equations of motion of our seven link planar biped using Lagrange method. In order to simulate the behavior of controlled system the Yobotics software is used, which is a free full featured software package written in Java language, develop in MIT leg’s lab for easily and quickly investigating the behavior of robots, biomechanical systems and mechanical devices. Keywords: Biped robot, Position/Force Control, Rough Terrain, Force Sensor, Yobotics, Trajectory Improvement
