Publications

This section provides a chronological overview of my publications. Fulltext papers and Bibtex information are provided for convenience.

Robust Stereo Visual Odometry Using Iterative Closest Multiple Lines

J. Witt, U. Weltin (IROS 2013)
Stereo Visual Odometry Using ICML

This work is concerned with the matching of straight lines between two stereo image pairs by reprojection. While we will focus on visual odometry in the realm of simultaneous mapping and localization, the techniques are also relevant to monocular and stereo 3D object detection and tracking. Our first contribution is an adaptation of the Iterative Closest Point (ICP) algorithm to the domain of lines. We argue that a naive "Iterative Closest Line" derivation cannot deliver similar performance. In contrast, our novel Iterative Closest Multiple Lines (ICML) algorithm allows efficient line matching while even reducing the amount of local minima during iterative optimization with its consideration of several weighted matches. The second contribution is a fast and robust hypothesize-and-test algorithm which can act as a fallback for challenging frame pairs where pure gradient-based optimization fails. In several differently textured scenes, we demonstrate robust performance, even in very sparse cases where proven feature point based methods fail. In comparison to edge-point ICP, we see speed improvements of more than an order of a magnitude and reduced susceptibility for local minima.

Bibtex Video

@inproceedings{Witt2013,
author = {Witt, J. and Weltin, U.},
title = {Robust Stereo Visual Odometry Using Iterative Closest Multiple Lines},
booktitle = {Proceedings of the IEEE/RSJ International Conference on Intelligent Robots and Systems},
pages = {XX--XX},
year = {2013},
organization={IEEE/RSJ}
}

Sparse Stereo by Edge-Based Search Using Dynamic Programming

J. Witt, U. Weltin (ICPR 2012)
Middlebury Tsukuba Edge Disparity Plot

In this paper, a novel edge-based stereo matching technique is presented. Depth discontinuities are specifically accounted for in the choice of support regions. We employ dynamic programming along edge-segments to efficiently enforce inter-scanline consistency. Although not performing a global optimization over the whole image we show that our approach performs successfully on the Middlebury benchmark datasets1 while being computationally feasible in real-time.

Bibtex

@inproceedings{Witt2012b,
author = {Witt, J. and Weltin, U.},
title = {Sparse Stereo by Edge-Based Search Using Dynamic Programming},
booktitle = {Proceedings of the 21st International Conference on Pattern Recognition},
pages = {3631--3635},
year = {2012},
organization={IEEE}
}

Robust Real-Time Stereo Edge Matching by Confidence-Based Refinement

J. Witt, U. Weltin (ICIRA 2012)
Middlebury Teddy Edge Disparity Plot

Edges are important features for tasks like object detection and vision-based navigation. In this paper, a novel real-time capable stereo edge refinement technique is presented. It propagates confidence and consistency along the detected edges, which reduces false matches significantly. Unmatched pixels are safely recovered by interpolation. We also investigate suitable support regions for edge-based matching. In the proposed solution, depth discontinuities are specifically accounted for. All approaches are extensively tested with the Middlebury benchmark datasets and compared to a sparse and several popular dense stereo algorithms.

Bibtex

@incollection{Witt2012a,
author = {Witt, J. and Weltin, U.},
title = {Robust Real-Time Stereo Edge Matching by Confidence-Based Refinement},
booktitle = {Intelligent Robotics and Applications},
series = {Lecture Notes in Computer Science},
editor = {Su, Chun-Yi and Rakheja, Subhash and Liu, Honghai},
publisher = {Springer},
pages = {512-522},
volume = {7508},
year = {2012}
}

Model Identification and H-∞ Attitude Control for Quadrotor MAV’s

O. Falkenberg, J. Witt, U. Pilz, U. Weltin, H. Werner (ICIRA 2012)
Quadrokopter H-infinity control step response

This paper presents the results of modelling, parameter identification and control of the rotational axes of a quadrotor robot. The modelling is done in Newton-Euler Formalism and has been published before. Contrarily, our method uses a Grey-Box-based, iterative parameter identification approach, the results of which can easily be reproduced and offers great accuracy. By neglecting nonlinear and cross-coupling effects, only three to four parameters have to be identified per axis, depending on the order of the motor dynamics. Based on the achieved results we were able to design an aggressive H-∞ attitude controller, which shows superior performance to the normal PID-like controllers. With an anti-windup compensator based on Riccati–equations we are able to show exceptional input disturbance rejection, even with disturbances saturating the engines.

Bibtex

@incollection{Falkenberg2012,
author = {Falkenberg, O. and Witt, J. and Pilz, U. and Weltin, U. and Werner, H.},
title = {Model Identification and \mathcal{H}_{\infty} Attitude Control for Quadrotor MAV’s},
booktitle = {Intelligent Robotics and Applications},
series = {Lecture Notes in Computer Science},
editor = {Su, Chun-Yi and Rakheja, Subhash and Liu, Honghai},
publisher = {Springer},
pages = {460-471},
volume = {7507},
year = {2012}
}

Quadrocopter Localization Using RTK-GPS and Vision-Based Trajectory Tracking

U. Pilz, W. Gropengießer, F. Walder, J. Witt, H. Werner (ICIRA 2011)
Quadrokopter System

This paper presents a method for trajectory tracking control and waypoint navigation of a quadrocopter using a vision-based camera system and an exact method for outdoor positioning using the Real-Time Kinematic Global Positioning System (RTK-GPS). First a quadrocopter prototype developed at the university is introduced. Then localization results of a RTK-GPS are presented, which show that the precision of absolute position data for the aerial vehicle can be improved significantly compared to a standard GPS solution. Experimental results of a simple PID position control scheme are presented which is used to suppress position errors of a trajectory tracking controller. The main result of this paper is the development of a trajectory tracking scheme in three dimensions which can be used to navigate to several waypoints along a predefined path. Both position controller and trajectory tracking controller experiments are conducted using data gathered from a Wii Remote camera system.

Bibtex

@incollection{Pilz2011,
author = {Pilz, U. and Gropengie{\ss}er, W. and Walder, F. and Witt, J. and Werner, H.},
title = {Quadrocopter Localization Using RTK-GPS and Vision-Based Trajectory Tracking},
booktitle = {Intelligent Robotics and Applications},
series = {Lecture Notes in Computer Science},
editor = {Jeschke, S. and Liu, H. and Schilberg, D.},
publisher = {Springer},
pages = {12-21},
volume = {7101},
year = {2011}
}

Design of a High Performance Quad-Rotor Robot Based on a Layered Real-Time System Architecture

J. Witt, B. Annighöfer, O. Falkenberg, U. Weltin (ICIRA 2011)
iQCopter

This work presents the results of an effort to create a high performance quad-rotor MAV named iQCopter for research in the field of swarm robotics and vision-based autonomous operation. During the design, navigational and computational capabilities have been of major priority. A distinctive feature of the iQCopter is its layered system architecture using a comparably powerful, hard real-time capable x86-computer even in the innermost control loop for maximum transparency and ease of use while a simple fixed-point microcontroller provides a low-level sensor interface and a fallback solution for safety reasons. Finally, a successful system identification and subsequent design of an aggressive H ∞ controller are presented as benchmark cases to demonstrate the performance of the design.

Bibtex

@incollection{Witt2011,
author = {Witt, J. and Annighöfer, B. and Falkenberg, O. and Weltin, U.},
title = {Design of a High Performance Quad-Rotor Robot Based on a Layered Real-Time System Architecture},
booktitle = {Intelligent Robotics and Applications},
series = {Lecture Notes in Computer Science},
editor = {Jeschke, S. and Liu, H. and Schilberg, D.},
publisher = {Springer},
pages = {312-323},
volume = {7101},
year = {2011}
}

Approximate Model Predictive Control for Nonlinear Multivariable Systems

J. Witt, H. Werner (Model Predictive Control, InTech)
Helicopter

The control of multi-input multi-output (MIMO) systems is a common problem in practical control scenarios. [...] If nonlinear dynamics are present in the plant, a linear model might not yield sufficient predictions for model predictive control techniques to function adequately. A related technique that can be applied to nonlinear plants is Approximate (Model) Predictive Control (APC). It uses an instantaneous linearization of a nonlinear model based on a neural network in each sampling instant. It is similar to generalized predictive control in most aspects except that the instantaneous linearization of the neural network yields an adaptive linear model. The main challenges in this work were the nonlinear, unstable and comparably fast dynamics of the 3-DOF helicopter. APC had to be extended to the MIMO case and model parameter filtering was proposed to achieve the desired control and disturbance rejection performance.

Bibtex

@incollection{Witt2010,
title = {Approximate Model Predictive Control for Nonlinear Multivariable Systems},
author = {Witt, J. and Werner, H.},
booktitle = {Model Predictive Control},
chapter = {6},
editor = {Zheng, Tao},
pages = {141--166},
publisher = {InTech},
year = {2010}
}

Go with the Flow: Optimal AUV Path Planning in Coastal Environments

J. Witt, M. Dunbabin (ACRA 2008)
Starbug

This paper describes a novel optimum path planning strategy for long duration AUV operations in environments with time-varying ocean currents. These currents can exceed the maximum achievable speed of the AUV, as well as temporally expose obstacles. In contrast to most other path planning strategies, paths have to be defined in time as well as space. The solution described here exploits ocean currents to achieve mission goals with minimal energy expenditure, or a tradeoff between mission time and required energy. The proposed algorithm uses a parallel swarm search as a means to reduce the susceptibility to large local minima on the complex cost surface. The performance of the optimisation algorithms is evaluated in simulation and experimentally with the Starbug AUV using a validated ocean model of Brisbane’s Moreton Bay.

Bibtex

@inproceedings{Witt2008,
title={Go with the Flow: Optimal AUV Path Planning in Coastal Environments},
author={Witt, J. and Dunbabin, M.},
booktitle={Proceedings of the Australian Conference on Robotics and Automation},
year={2008}
}

Approximate Model Predictive Control of a 3-DOF Helicopter

J. Witt, S. Boonto, H. Werner (CDC 2007)
Helicopter

This paper presents the design and experimental validation of a nonlinear multivariable predictive controller for an educational 3-DOF helicopter system. The control strategy - Approximate Predictive Control (APC) - is based on a neural network model of the nonlinear plant and its linearization in each sampling instant. The control input is generated using the linearized model and a GPC control law. For this application, a previously published APC approach has been extended to MIMO systems, and was modified by introducing model parameter filtering to achieve the required control performance. Experimental results demonstrate good tracking and disturbance rejection performance of the proposed method.

Bibtex

@inproceedings{Witt2007,
title={Approximate Model Predictive Control of a 3-DOF Helicopter},
author={Witt, J. and Boonto, S. and Werner, H.},
booktitle={Proceedings of the 46th IEEE Conference on Decision and Control},
pages={4501--4506},
year={2007},
organization={IEEE}
}