- Master Thesis
Rights / licenseIn Copyright - Non-Commercial Use Permitted
Fast, ecient and versatile locomotion is the objective of the cutting-edge research of legged robots. Researchers are faced with two basic questions: First, what kind of motion is required to move the robot fast and ecient? Second, how has the robot to be controlled to achieve this motion? A well-known approach in the research area is to characterize the demanded motion by a so-called template, a simplied physical model that generates the same motion as observed in the locomotion of animals and humans. Whereby the anchor, a more complex model of the musculoskeletal system of the vertebrates, is performing the same basic motion as the template . Biomechanic studies showed that dynamic locomotion, especially the simplest version of single legged hopping, is the Spring Loaded Inverted Pendulum model that describes the motion of the center of gravity of the whole body. The same motion can be enforced with the anchor by a control design based on the operational space formulation. The aim of this thesis is to design and implement a controller for the Series Compliant Articulated Robotic Leg of ETH to accomplish template-based hopping in simulation and experiments. A new framework is written within the software package SL  that facilitates the implementation and evaluation of the sophisticated controller. The implemented software architecture includes a low-level controller on an embedded system that allows position and torque control. The existing low-level framework is optimized and improved so that a fast communication between the control layers becomes possible. The controller successfully demonstrates template-based hopping in simulations, but fails in experiments. Parts of the controller are experimentally veried with success and the problem is identied as a hardware problem Show more
External linksFull text via SFX
PublisherEidgenössische Technische Hochschule Zürich, Autonomous Systems Lab
SubjectCOMPUTER VISION + SCENE UNDERSTANDING (ARTIFICIAL INTELLIGENCE); COMPUTERVISION (KÜNSTLICHE INTELLIGENZ); PROGRAMS AND ALGORITHMS FOR THE SOLUTION OF SPECIAL PROBLEMS; ROBOTERNAVIGATION; MOBILE ROBOTS; PROGRAMME UND ALGORITHMEN ZUR LÖSUNG SPEZIELLER PROBLEME; ROBOT NAVIGATION; MOBILE ROBOTER
Organisational unit03737 - Siegwart, Roland Y.
MoreShow all metadata