| This thesis focuses on offline running synthesis and online running control designs for robots with one and two articulated legs. Since no compliant parts are used in the hardware designs, the robots are hard to stabilize due to large ground reaction forces. Study of extreme configurations may provide acute understandings of legged locomotion systems.A finite-time controller is employed in the flight phase to improve landing accuracy. In the stance phase, the controller is composed of three modules. The finite-time position-tracking module, designed with the same principle as the flight controller, prepares correct initial state for the subsequent flight phase. The force-suppression module rejects excessive external forces, preventing robot damage. The online ZMP compensator drags the ZMP closer to the center of the support range, with sacrifice of position tracking accuracy, and thus, the running stability can be sustained.Simulations have demonstrated the effectiveness of the proposed approaches.Inspired by new results in biological sciences, biomechanical analysis, and legged robotics, a fundamental assumption is made: the energy cost of the robot in the flight phase is small, when the robot runs on flat even ground. This assumption is formulated as a static optimization problem. Solving this static optimization problem produces the initial joint velocities for the flight phase. The running gaits can then be generated by dynamic optimization. The ground reaction forces are constrained within the permitted range. The stability criterion based on the Zero-Moment Point (ZMP) serves as other nonlinear constraints. |
