We present a method for controlling the forward speed and the apex height of a one-legged hopping robot over rough terrain, using a single actuator located at the robot hip. The control algorithm is comprised of two elements, the forward speed control and the height control. The only input to the system is the torque applied by the hip actuator. The control is demonstrated to perform tracking of desired forward speed trajectories and desired apex height trajectories. Simulation and experimental results on the SAHR (Single Actuator Hopping Robot) experimental setup are presented and compared. It is shown that the robot follows both trajectories closely in simulation as well as in experiment. Also the robot is tested successfully on a rough terrain course, which includes inclined ground and an abrupt drop in height of over 25% the length of the robot leg. The robot has no knowledge of its environment. Further, the robot is made to run over the course a number of times, to demonstrate the control robustness.

In this paper a novel multipart control is developed for a trotting quadruped robot. The control is designed to drive the quadruped to a steady-state motion with desired forward speed and apex height, using only one actuator per leg. The body pitching motion is controlled to be small. The controller is applied to the robot and the complete system is used to develop a parametric study for the robot. The study examines the behavior of the actuator effort and the leg touchdown angles, over a parametric region of both the robot physical parameters and the gait parameters. Interesting results appear, not previously reported, that may contribute to enhanced robot design and better gait selection for a given robot. Typical findings are that a robot should be lighter- weight when running on more slippery terrain, as well as that certain higher forward speeds require less actuator effort than other slower speeds.

In this work a novel control method is presented for controlling the forward speed and apex height of a special class of running quadruped robot, with a dimensionless inertia of 1, and one actuator per leg. Seeking to minimize the parasitic pitching motion in running, pronking is used as the target gait. The control design is based on the robot dynamics, allowing its application to a wide range of robots of the class studied. Moreover, the controller adjusts the robot speed and height, requiring knowledge only of the robot physical parameters. The control ensures that negative actuator work during the stance phase is zero, thereby reducing the power expenditure. Small, off-the-shelf DC motors are adequate for the control implementation, while results of application to a detailed robot model show good performance even when including leg mass, foot collision, motor limitations, foot slipping and other factors.

In this paper it is shown that, for a one-legged robot, there exists a particular passive gait, of all those possible, for which the dissipated energy per meter of travel is minimized. An analytical method is used to identify the optimal gait. A SLIP model of the robot is used to simplify the dynamics. Both mechanical and electrical losses are taken into account. A numerical analysis of a complete robot model follows, to evaluate the accuracy of the analytical prediction. Finally, the limitations imposed by a torque limited motor, with regard to the optimal gait, are studied.