Design of a running robot and the effects of foot placement in the transverse plane

Timothy Sullivan, Purdue University

Abstract

The purpose of this thesis is to make advances in the design of humanoid bipedal running robots. We focus on achieving dynamic running locomotion because it is one metric by which we can measure how far robotic technologies have advanced, in relation to existing benchmarks set by humans and other animals. Designing a running human-inspired robot is challenging because human bodies are exceptionally complex mechanisms to mimic. There are only a few humanoid robots designed specifically for running and the existing robots are either constrained to a plane, do not yet exhibit human-like motion, or are unstable. One aspect of bipedal running dynamics that could be understood better, and could improve the performance of these robots, is the role of foot placement in the transverse plane, via rotation of the hip, in the sagittal and coronal planes. To achieve this objective, I have developed and studied a running simulation, designed and built a bipedal running robot, and then conducted an experimental study on the effects of step width on overall whole-body locomotion dynamics. The planar running simulation investigates foot placement in the sagittal plane, by changing the angle of the hip at the beginning of each stance phase. The model effectively mimics a human-like response to a horizontal or vertical velocity disturbance while running. This is achieved by implementing a controller which changes the hip angle at touch down and by using the energy of the system as control feedback. This controller effectively reduces the time required to recover from a perturbation, reduces the overall energy expended recovering from a perturbation, and allows the legged model to function at significantly lower torque values. The bipedal running robot is designed as a research platform for systematic investigation of key mechanisms and control methods, in order to progressively move towards a more biologically accurate and stable running robot. The result of this design is a simple, lightweight, and powerful research platform which can easily be expanded on in order to systematically study the effects of individual parameters. The step width study investigates the effects of step width in the coronal plane on a 3D 1DOF per leg bipedal running gait. The results of this study help to confirm biological hypothesis on how stable running is achieved and suggests the next steps in moving towards a more biologically correct 3D running robot. A range of step widths were found for which stable locomotion existed, where stability is empirically assessed by checking if a run surpasses 20 steps. The smallest step width studied exhibited unstable rolling behavior, which confirms previous theory stating that dimensionless inertia, a function of step width and mass moment of inertia, effects roll stability. Similarly, yaw stability was observed to increase with wider step widths. The inability to stabilize roll and yaw for narrow step widths is consistent with biological studies where roll and yaw stability are thought to be mainly achieved by active use of the upper body. The widest step width studied exhibited unstable behavior in pitch. Pitching could be significantly impacted by lateral ground reaction forces, the distance from the hip to the center of mass, and changes in the relative magnitudes of roll, pitch, and yaw inertias. It was found that lateral ground reaction forces increase and the vertical distance from hip to center of mass decreases with increasing step width. Further, the magnitude of yaw inertia gets closer to that of pitch. The convergence of these factors suggest that a presently unidentified coupling in roll, pitch, and yaw occurs that yields the observed behavior for wide step widths.^

Degree

M.S.M.E.

Advisors

Justin Seipel, Purdue University.

Subject Area

Engineering, Robotics

COinS