Physical models raise questions that computer-models do not pose. In this project experimental biomechanical findings, neural control concepts and computational model predictions are biomimetically translated into robots to proof concepts, raise new questions and to better understand human leg function.
Involved People: Christophe Maufroy, Moritz Maus, Martin Grimmer, Daniel Maykranz, André Seyfarth
Previous people: James Andrew Smith, Fumiya Iida, Poramate Manoonpong, Daniel Renjewski
JenaWalker is a series of bipedal robots having three-segmented legs. The segments are connected via mono- and biarticular springs or elastic tendons. The JenaWalker is basically a testbed for investigating passive structures and neural control required for human-like walking on elastic legs.
RunBot is bipedal robot developed at the Bernstein Center for Computational Neuroscience Göttingen. It is based on the principle of passive dynamic walkers having extended and therewith stiff legs during stance phase. The trunk is kept upright using a neural network and novel learning strategies.
The fundamental model for investigating bipedal gaits is the spring-mass model. In this template the legs are represented as massless springs and the body is reduced to a point mass. The spring-mass model is able to show the leg dynamics in both, walking and running.
Numerous biomechanical variables are supplied by a number of experiments on human walking and running. These include ground reaction forces, motion analysis, and surface EMG, which are basis for a broad variety of analysis.
The JenaWalkerII robot was built to investigate the role of elastic structures in the leg during human locomotion. By introducing simple sinusoidal hip oscillations (DC motors), walking, hopping and running movements can be observed. No sensory feedback was required to stabilize locomotion.