Khadiv, MajidMajidKhadiv0000-0001-9889-6543Khorshidi, ShahramShahramKhorshidi2025-05-212025-05-212022Technische Universität Hamburg (2022)https://hdl.handle.net/11420/47725Legged robots require knowledge of the position, velocity and orientation of the base in order to maintain stability and execute controlling schemes. We present a state estimation framework for legged robots locomotion based on Extended Kalman Filtering (EKF). The core idea of this approach is to exploit the information from the kinematics constraints given by the feet in contact with the ground at each interval, and then fuse it with the sensory data from Inertial Measurement Unit (IMU) on the robot, therefore the whole task of control is executed by the information provided with onboard sensory devices. Consequently, we will design the estimator (EKF) based on the formulation of the underlying stochastic model. As shown in previous works, such a filter is observable for the linear velocity vector, and roll and pitch angles of the base, which are the crucial elements in order to use the estimated states for the feedback to the whole body controllers in legged robots locomotion. To this end, the estimator is implemented and tested with different types of motion on the quadruped robot Solo12, and the correctness and reliability of such a framework is shown. Finally, the observer-based controller is successfully tested in a feedback scheme for the whole body control of the robot without using any external sensory device.enhttps://creativecommons.org/licenses/by/4.0/Legged robots | State estimation | Extended Kalman filter | Control | Solo12Technology::629: Other Branches::629.8: Control and Feedback Control SystemsComputer Science, Information and General Works::006: Special computer methodsMaster Thesishttps://doi.org/10.15480/882.963710.15480/882.9637Seifried, RobertRobertSeifriedHoffmann, NorbertNorbertHoffmannOther