Michael W. Hofbaur

Mode Estimation of Probabilistic Hybrid Systems

Model-based diagnosis and mode estimation capabilities excel at diagnosing systems whose symptoms are clearly distinguished from normal behavior. A strength of mode estimation, in particular, is its ability to track a systems discrete dynamics as it moves between different behavioral modes. However, often failures bury their symptoms amongst the signal noise, until their effects become catastrophic.We introduce a hybrid mode estimation system that extracts mode estimates from subtle symptoms. First, we introduce a modeling formalism, called concurrent probabilistic hybrid automata (cPHA), that merge hidden Markov models (HMM) with continuous dynamical system models. Second, we introduce hybrid estimation as a method for tracking and diagnosing cPHA, by unifying traditional continuous state observers with HMM belief update. Finally, we introduce a novel, any-time, any-space algorithm for computing approximate hybrid estimates.

Model-based programming of fault-aware systems

A wide range of sensor-rich, networked embedded systems are being created that must operate robustly for years in the face of novel failures by managing complex autonomic processes. These systems are being composed, for example, into vast networks of space, air, ground, and underwater vehicles. Our objective is to revolutionize the way in which we control these new artifacts by creating reactive model-based programming languages that enable everyday systems to reason intelligently and enable machines to explore other worlds. A model-based program is state and fault aware; it elevates the programming task to specifying intended state evolutions of a system. The programs executive automatically coordinates system interactions to achieve these states, entertaining known and potential failures, using models of its constituents and environment. At the executives core is a method, called CONFLICT-DIRECTED A*, which quickly prunes promising but infeasible solutions, using a form of one-shot learning. This approach has been demonstrated on a range of systems, including the National Aeronautics and Space Administrations Deep Space One probe. Model-based programming is being generalized to hybrid discrete-continuous systems and the coordination of networks of robotic vehicles.

Improving Robustness of Mobile Robots Using Model-based Reasoning

Retaining functionality of a mobile robot in the presence of faults is of particular interest in autonomous robotics. From our experiences in robotics we know that hardware is one of the weak points in mobile robots. In this paper we present the foundations of a system that automatically monitors the driving device of a mobile robot. In case of a detected fault, e.g., a broken motor, the system automatically reconfigures the robot in order to still allow to reach a certain position. The described system is based on a generalized model of the motion hardware. High-level control like path-planner only to change its behavior in case of a serious damage. The high-level control system remains the same. In the paper we present the model and the foundations of the diagnosis and reconfiguration system.

Model-based fault diagnosis and reconfiguration of robot drives

Modern drives of mobile robots are complex machines. Because of this complexity, as well as of wear and aging of components, faults occurs in such systems quite frequently at runtime. In order to use such drives in truly autonomous robots it is desirable that the robot is able to automatically react to such faults. Therefore, the robot needs reasoning and reconfiguration capabilities in order to be able to detect, localize and repair such faults on-line. In this paper we propose a model-based diagnosis and reconfiguration framework which allows an autonomous robot to detect and compensate faults in its drive. Moreover, we present an implementation for a real robot platform. Finally, we report experimental results which shows that the proposed framework is able to correctly cope with injected faults in the drive hardware, like broken motors.

Modular re-configurable robot drives

We propose a modular platform for wheeled mobile robots that utilises a 6-edge honey-comb prism as its basic building block to realize robot drives of diverse geometry. In terms of functionality, we designed a specific wheel suspension for a drive-module comb that can utilise both, a standard wheel or a Mecanum wheel. A quick-lock interconnection mechanism for the comb modules allows us to quickly configure/reconfigure various robot drives and enables us to realise autonomous wheeled robots with the ability to connect to other robots or even to reconfigure the robots geometry. This configuration capability offers many interesting opportunities for robotics research since we can adapt a robot in terms of its kinematic functionality, payload and size.

Sensor selection for fault parameter identification applied to an internal combustion engine

This paper presents a methodology for the selection of suitable sensors for a diagnosis problem that is relevant for control of internal combustion engines. The diagnosis is formulated as a parameter estimation problem, which is solved by augmenting the plant model with additional states that represent the faults. An Extended Kalman Filter is applied to estimate the states of the augmented model. The sensor selection is based on the classical concepts of observability as well as on the distinguishability analysis for the faults.

Online trajectory optimization for nonlinear systems by the concept of a model control loop — Applied to the reaction wheel pendulum

This paper presents a solution for online optimization of reference trajectories for nonlinear tracking controllers. A Nonlinear Model Predictive Control (NMPC) algorithm is applied to control a model of the plant online to track the user references, e.g. step changes. The resulting state trajectories of the simulated model are used as a reference for a low level trajectory tracking controller that regulates the plant. Additionally, the control signals of this simulated closed loop can be used as feedforward control in the low level controller. This concept is called “model control loop”, that is a special form of a two-degree-of-freedom control. In this setup, the advantage of online trajectory optimization that NMPC offers is exploited, while the stability problem of NMPC due to long computation times is avoided. The method is applied to the reaction wheel pendulum, experimental results are presented.

An interleaved, model-supported system identification scheme for the particle accelerator CLIC

The particle accelerator CLIC is a future linear collider, which is developed at CERN. The quality of the particle-beams produced by CLIC is very sensitive to ground motion. The efficiency of the feedback used to counteract ground motion, relies crucially on the quality of the system knowledge. Therefore, we present a system identification scheme to follow changes of accelerator parameters. The algorithm is based on the well-known RLS (recursive least squares) algorithm with exponential forgetting, but adds modifications to improve the learning speed and to address excitation on-strains given by the system. Parallel-running, interleaved RLS algorithms identify parts of the overall system. The different results are combined by using a priori knowledge. Parts that could not be identified directly, are extrapolated with the help of physical models. The modified algorithm can follow system changes with a factor of approx. 30 improved learning speed, compared to the conventional RLS algorithm. It works robustly, in spite of sensor noise and disturbances acting on the excitation signals. The prize that has to be paid is a minimum permanent error of 13% due to model errors. The scheme can easily be adapted to other linear accelerators. Moreover it should be possible to reduce the steady-state error of the identification for other machines, since the main linac of CLIC is an especially difficult system to model and excite.

Redundancy Resolution of a 9 DOF Serial Manipulator Under Hard Task Constraints

Kinematic redundancy is a topic which has been discussed under various contexts to co-achieve a primary task along with additional tasks. However a structural analysis about kinematic redundancy to comprehend what really is happening underneath the hood has seldom been a topic of discussion. In this paper, kinematic redundancy is viewed from a slightly different perspective to deduce the constraints in the null space. Task distortion is explained in a geometrical sense and it lays the foundation for possible corrective measures for a better realization of the task. For the sake of generality, the considered primary task in this work is highly constraining in nature. A 9 DOF serial manipulator will be used instead of a conventional 7 DOF manipulator as the constraints imposed by the primary task results in insufficient null space dimensions.

Practically stabilizing motion control of mobile robots with steering wheels

In this work, a novel approach for control of wheeled mobile robots (WMRs) with center-steerable standard wheels is presented. The method bases on static input-output linearization and leads to a control law that can be employed for tracking of reference postures and practical stabilization of rest postures. The controller does neither neglect any physical constraints nor introduce new singularities into the closed loop system. Its practicability is demonstrated in simulation and real-world experiments.