Marc Hildebrandt

Computer-Based Control of Deep-Sea Manipulators

Traditionally, control of deep-sea manipulators happens in a master-slave fashion. The resulting necessity of a specially trained operator on-site severely limits the disposability of such systems. Further even simple tasks require significant amounts of time due to a number of factors, e.g. the operators limited view of the manipulation environment or the low level of intuitive sensory feedback of a manipulators actions. The CManipulator project of the DFKI-Lab Bremen, Germany, addresses these problems by utilising state-of-the-art computing algorithms to allow the partial automation of common tasks in deep-sea manipulation. This paper focuses on the basic principles and problems of such control structures on the example of the widely used hydraulic Orion 7P by Schilling Robotics.

IMU-aided stereo visual odometry for ground-tracking AUV applications

This paper addresses the problem of AUV navigation by showing the feasibility of a stereo visual-inertial approach to odometry retrieval for an AUV. This information is intended as input for a complete SLAM system. After its classification among many other similar approaches in recent work is shown, the algorithm is described in detail. A number of experiments conducted on synthetic data show the performance in respect to precision and computational cost. As a conclusion, future extensions and applications are briefly discussed.

Realtime motion compensation for ROV-based tele-operated underwater manipulators

This paper presents a novel underwater movement compensation algorithm for stabilization of manipulator position utilizing not ROV movements for disturbance comensation, but overlaid manipulator movements. A model based estimator is used to predict vehicle movement and provide the manipulation system with the necessary time to compensate for the estimated motion. It describes the conceptual benefits of this approach compared with common station-keeping algorithms, and shows how previous methods can be combined with the new approach in order to further improve manipulator position accuracy. The method is validated in a number of experiments, which show its feasability and outstanding performance.

Adaptive AUV mission management in under-informed situations

Autonomous Underwater Vehicles (AUVs) are in high demand within the offshore industry and maritime research, mainly used for bathymetry and data acquisition. The control architectures of these AUVs mimic this primary function by focusing on strict mission plans as these kind of application require, thus reducing the need for direct sensor reaction to emergency situations. The emerging needs for more complex underwater application like the inspection of structures, search missions or taking samples from the floor or in the water column with respect to certain environmental conditions demand more adaptive, currently not existing, control architectures. The main problem hereby is that, opposed to non-underwater application scenarios for autonomous systems, the lack of a stable communication channel to the vehicle demands complete autonomy. The architecture proposed in this paper aims at tackling the issue of unpredictability. The main issue, especially in exploration or inspection missions, is that little is known at the beginning of the mission. This lack of information makes planning meaningless, as the planner has no idea whatsoever as to what should be done while on site. Our proposed architecture offers to replace, in these under-informed situations, planning-based approaches by a plan management approach. This approach is able to use both predictive (planning) approaches and behaviours (reactive) approaches to control the system, which is then used to execute and control execution of functional components. The mixing of these decision-making schemes being done based on the information available to the system. This paper presents the general idea of our architecture as well as the implementation and a validation experiment with the AUV AVALON.

Two years of experiments with the AUV dagon - a versatile vehicle for high precision visual mapping and algorithm evaluation

The AUV DAGON was designed as a vehicle for algorithm evaluation and visual mapping. Since its initial launching in early 2010 two years of experiments and experience with the vehicle have passed. This paper will give an overview of the work with the AUV DAGON and highlight the scientific experiments conducted with it, concluding with a “lessons learned” section with important modifications and ideas for future vehicles.

Design and test of a robust docking system for hovering AUVs

This paper presents the conceptual idea and a developed prototype of a docking system for a hovering AUV (Autonomous Underwater Vehicle). The presented docking system passively assists the vehicles docking procedure by guiding it into a defined position, where it is mechanically locked in place by one of the stations actuators. In its final position the docking system provides an electrical power supply and a broadband data link to the AUV. The developed concept additionally enables differently shaped AUVs to use the same docking station.

Hardware ROV simulation facility for the evaluation of novel underwater manipulation techniques

In this paper we describe a hardware facility to simulate movements of a remotely operated underwater vehicle (ROV) in a water basin for the evaluation of novel underwater manipulation techniques as it was build at the underwater robotics department of the German Research Center for Artificial Intelligence (DFKI) in Bremen. The three main functionalities of the ROV simulator are to drive predefined static trajectories (e.g. recorded from real ROV mission movements), to virtually generate realistic ROV movements to approximate the desired ROV behavior and to let the system react in a realistic way to forces emerging while one of the attached manipulators interacts with an object. For the realistic ROV movement generation we are using a vectorial model representation of the simulated ROV based on established dynamic equations of motion for six degrees of freedom.

A practical underwater 3D-Laserscanner

A number of attempts have been made to use the benefits of 3D-Laserscanning techniques in the underwater environment. Unfortunately, due to a number of operative problems with such devices, their accuracy and therefore applicability remains quite low. This paper specifically focuses on these practical issues by expanding on previous works in this area and improving their usability. The result is a calibration procedure for triangulation-based 3D-laserscanners for the underwater environment which provides a very promising precision and reliability, but at the same time does not demand exaggerated deployment overhead.

A Validation Process for Underwater Localization Algorithms

This paper describes the validation process of a localization algorithm for underwater vehicles. In order to develop new localization algorithms, it is essential to characterize them with regard to their accuracy, long-term stability and robustness to external sources of noise. This is only possible if a gold-standard reference localization (GSRL) is available against which any new localization algorithm (NLA) can be tested. This process requires a vehicle which carries all the required sensor and processing systems for both the GSRL and the NLA. This paper will show the necessity of such a validation process, briefly sketch the test vehicle and its capabilities, describe the challenges in computing the localizations of both the GSRL and the NLA simultaneously for comparison, and conclude with experimental data of real-world trials.

Experimental evaluation of various machine learning regression methods for model identification of autonomous underwater vehicles

In this work we investigate the identification of a motion model for an autonomous underwater vehicle by applying different machine learning (ML) regression methods. By using the data collected from the robots on-board navigation sensors, we train the regression models to learn the damping term which is regarded as one of the most uncertain components of the motion model. Four regression techniques are investigated namely, artificial neural networks, support vector machines, kernel ridge regression, and Gaussian processes regression. The performance of the identified models is tested through real experimental scenarios performed with the AUV Leng. The novelty of this work is the identification of an underwater vehicles motion model, for the first time, through machine learning methods by using the robots onboard sensory data. Results show that the damping model learned with nonlinear methods yield better estimates than the simplified linear and quadratic model which is identified with least-squares technique.