PUB - Publikationen an der Universität Bielefeld

The term "biorobotics" denotes a field in which biologically inspired robotic implementations are designed and evaluated. Through a "learning from nature", engineers hope to solve robotic problems with the efficient solutions that biology has found. Several of the biologically inspired algorithms are based on visual input and need a distinct computational performance. In mobile robotics, however, small and low-power solutions are needed. A fairly well trade-off with a relatively low power consumption and a relatively high computational performance can be achieved by application-specific hardware-based designs. In this thesis, four hardware-based and biologically inspired implementations are described: - One analog and one digital (FPGA-based) implementation of a motion detection model which was derived from results of behavioral experiments with beetles. - Two digital implementations of local navigation algorithms that were proposed based on the so-called "snapshot model". The "snapshot model" gives an explanation for the navigational capabilities of insects like bees. Besides the description of the four hardware-based implementations, this dissertation includes introductory sections which explain the biological models and algorithms in general, outline central parts of the insect visual system, give an overview of different ways of hardware-based implementation, and show four typical robotic setups. Finally, overall discussion and conclusion chapters compare the presented four designs, discuss advantages and drawbacks and give an outlook on future work. The appendix includes an overview of current analog and mixed-signal configurable integrated circuits.