Design, simulation, and stability of a hexapedal running robot

The ability of animals to navigate rough terrain is currently unmatched by their man-made counterparts. Recent studies by biologists have begun to demonstrate some of the principles behind their remarkable capabilities. In particular, studies of cockroaches have shown that they use a feed-forward motor actuation pattern and considerable structural compliance to run quickly over very rough terrain. Their sprawled posture and tuned impedance in their musculoskeletal system enable an instantaneous or 'preflex' response to disturbances. This allows for rapid response to the large perturbations experienced when interacting with irregular terrain.; Consideration of these principles has led to the design of the Sprawl family of robots, which features one active thruster and one entirely passive rotary joint on each leg. Without these spring elements the robots would not be able to run. With them, they can easily overcome hip-height obstacles without any alteration of their open-loop controller.; This thesis describes the development, calibration, and analysis of a three-dimensional dynamic simulation of the Sprawl robots. This simulation was developed as a design tool to investigate the effects of parameter variation, and to gain understanding about how to tune the leg configuration and hip impedance which constitute the self-stabilizing posture of the robot.; The simulation is used to characterize the sensitivity of the system's performance to changes in the robots' physical parameters. The key parameters that influence speed and stability are identified, and their effects and the nature of their coupling are investigated.; While speed is easy to measure, no universal metric for running stability exists. This thesis discusses four distinct steady-state measures of stability that are applicable to a simulated running robot. The effect of modifying the posture of the robot on stability is investigated for each of these measures.; As a demonstration of its utility as a design tool, the simulation is used to tune the performance of one of the Sprawl robots. Changing the leg design according to the model's predictions resulted in a two-fold increase of the robot's speed.