Goals to be achieved
The Cognitive Neuroinformatics group contributes to TRIPLE-nanoAUV by developing software components for the nanoAUV, a miniaturized autonomous underwater vehicle, that enables the system to autonomously plan and execute its mission. The main objectives are to develop strategies for locating regions of interest, perform mission and motion planning to survey these areas, and while doing so, cope with potentially noisy sensor measurements and uncertain information. The main goal is to allow the nanoAUV to successfully complete its mission and to safely return to the ice shuttle.
Tasks within the project
- Conception, design, and implementation of a subsystem for autonomous decision making that enables the nanoAUV to accomplishing its main mission as well as potential secondary objectives.
- Identify mandatory information required as a foundation for decision making which includes a priori knowledge as well as information available during mission execution (e.g. provided by sensor fusion).
- Explicitly model incomplete and uncertain information to be used in the decision making process.
- Develop optimal search strategies for the given scenario, aiming for a maximization of information gain while simultaneously ensuring system safety and localization accuracy.
- Investigate autonomously executable fallback strategies and behaviors to be executed in case of errors.
Preliminary work
One research focus of the Cognitive Neuroinformatics group is the representation and handling of uncertain, incomplete, and inconsistent knowledge from a theoretically point of view as well as in multiple, practically relevant problems and applications. This includes but is not limited to localization, mapping, planning, and decision making for different autonomous systems, especially in autonomous driving and space robotics. With respect to TRIPLE-nanoAUV, the DLR projects KaNaRiA and KaNaRiA-K²I as well as Enceladus Explorer (EnEx) and EnEx-CAUSE are the most relevant ones. The KaNaRiA projects focus on autonomous navigation in the vicinity as well as on the surface of a small celestial body, which covers spacecraft and rover operations. The goal of the EnEx projects is the development of novel autonomous navigation techniques for in-ice exploration of glaciers and icy bodies of the Solar System using multiple, cooperative melting probes. This was successfully demonstrated during multiple field tests on glaciers in the Alps and Antarctica.
Implementation steps
Starting out with a concurrent engineering design study, the TRIPLE-nanoAUV project entered its concept design and evaluation stage. With respect the autonomous decision making module, the focus is on the following tasks:
- Development of a complex, autonomous decision making approach considering resource limitations. This includes limited computational power and memory as well as reduced availability of information because of miniaturized and a limited number of sensors.
- Achievement of the required robustness for the planned mission without or only with minimal support from an operator.
- Development of a simulation system for the overall nanoAUV mission with a reasonable tradeoff between complexity, simulation depth, and abstraction.
- Evaluation and validation of the nanoAUV GNC system, in particular the autonomous decision making module, under varying external and internal parameters in simulation in order to improve the capabilities for the deployment in a harsh environment with partially unknown conditions.
- Demonstration of the autonomous mission execution in an analogous environment, first with a suitable off-the-shelf AUV and later with the actual nanoAUV.
Points of Contact
Joachim Clemens, clemens@uni-bremen.de
Carsten Rachuy, rachuy@uni-bremen.de
Till Koch, tiko@uni-bremen.de