Tim Braun

Design Space Exploration for the BLAST Algorithm Implementation

The design and implementing of a key point detector on embedded reconfigurable hardware is investigated. The major challenges are efficient hardware/software partitioning of the key point detector algorithm, data flow management as well as efficient use of memory, bus and processor. We present a modular and manual hardware/software co-design, with its implementation on a Xilinx XUP-Virtex II Pro board co-design to solve these issues.This paper proposes optimizations of the methods and parameters used in both mathematical approximation and hardware design for logarithmic number system (LNS) arithmetic. First, we introduce a general polynomial approximation approach with an adaptive divide-in-halves segmentation method for evaluation of LNS arithmetic functions. Second, we develop a library generator that automatically generates optimized LNS arithmetic units with a wide bit-width range from 21 to 64 bits, to support LNS application development and design exploration. The basic arithmetic units are tested on practical FPGA boards as well as software simulation. When compared with existing LNS designs, our generated units provide in most cases 6% to 37% reduction in area and 20% to 50% reduction in latency. The key challenge for LNS remains on the application level. We show the performance of LNS versus floating-point for realistic applications: digital sine/cosine waveform generator, matrix multiplication and radiative Monte Carlo simulation. Our infrastructure for fast prototyping LNS FPGA applications allows us to efficiently study LNS number representation and its tradeoffs in speed and size when compared with floating-point designs.

{RAVON} -- The Robust Autonomous Vehicle for Off-road Navigation

Abstract: This chapter describes the work of the Robotics Research Lab at the University of Kaiserslautern in the field of autonomous off-road robotics. It introduces concepts developed for hazard detection, terrain classification, and collision-free autonomous navigation. As an example of a system implementing the described techniques, the mobile off-road robot RAVON is presented. Experiments have been carried out to prove the effectiveness of the approaches.

Probabilistic distance measures of the Dirichlet and Beta distributions

We give the analytical definitions of the Chernoff, Bhattacharyya and Jeffreys-Matusita probabilistic distances between two Dirichlet distributions and two Beta distributions as its special case. For all other known probabilistic distances we show their inappropriateness in the analytical case. We discuss the parameter learning of the Dirichlet distribution from a finite sample set and present an application for split-and-merge image segmentation.

A Customizable, Multi-Host Simulation and Visualization Framework for Robot Applications

A highly flexible framework for visualization and sensor simulation in three-dimensional environments is presented. By allowing the insertion of freely programmable elements for online scene modification, a programmer can customize the framework to fulfill the exact simulation or visualization needs of an application of interest. Furthermore, the framework provides simple external interfaces so that multiple clients can be attached to it with ease. The frameworks’ capabilities are demonstrated with two complex robotic applications that require both a high quality simulation of cameras and lasers scanners and an intuitive 3D visualization.

Visual Terrain Traversability Estimation Using a Combined Slope/Elevation Model

A stereo vision based terrain traversability estimation method for offroad mobile robots is presented. The method models surrounding terrain using either sloped planes or a digital elevation model, based on the availability of suitable input data. This combination of two surface modeling techniques increases range and information content of the resulting terrain map.

Emotion Based Control Architecture for Robotics Applications

Assistance and service systems are one of the main research topics in robotics today. A major problem for creating these systems is that they have to work and navigate in the real world. Because this world is too complex to model, these robots need to make intelligent decisions and create an intelligent behavior without knowing everything about the current situation. For these aspects, the importance of emotion increases, because the emotional influence helps human beings as well as animals to make their decisions. To enable a robot to use emotions, a concept for an emotion based control architecture was designed. The basis of this architecture is a behavior based approach. This paper presents the developed architecture. Furthermore two application possibilities are presented, where parts of the architecture were already tested and implemented.

Cost-Efficient Global Robot Navigation in Rugged Off-Road Terrain

The thesis “Cost-Efficient Global Robot Navigation in Rugged Off-Road Terrain” proposes a global robot navigation concept for rugged off-road terrain which is robust against inaccurate self-localization and scalable to large environments, but also cost-conscious, e.g. able to generate navigation paths which optimize a cost measure closely related to terrain traversability. The thesis contributes a set of new techniques to integrate cost-consciousness into a primarily topological navigation scheme. Also, novel methods to learn and optimize cost estimates from experience are developed.

SoPC Architecture for a Key Point Detector

The design and implementing of a key point detector on embedded reconfigurable hardware is investigated. The major challenges are efficient hardware/software partitioning of the key point detector algorithm, data flow management as well as efficient use of memory, bus and processor. We present a modular and manual hardware/software co-design, with its implementation on a Xilinx XUP-Virtex II Pro board to solve these issues.

Bhattacharyya probabilistic distance of the Dirichlet density and its application to Split-and-Merge image segmentation

We derive the Bhattacharyya distance between two Dirichlet densities. As an application we use image segmentation by a split-and-merge algorithm.

A Humanoid Head for Assistance Robots

For the development of a humanoid robot able to assist humans in their daily life, adequate interaction is a key feature. If one considers that in human communication more than 60% is non verbal (communication with facial expressions and gestures) an important research topic is how should a robot head look like and what skills must such a head have to be able to interact with humans. In the following first the simulation system is introduce which was used to test mimic and the expression of emotion. Based on this simulatiuon results the mechatronical head of the University of Kaiserslautern is designed. Finaly a real-time capable method for image based face detection is presented, which is a basic ability needed for interaction with humans.