@inproceedings{clemens2018autonomous,
	author = {Joachim Clemens and Christian Meerpohl and Verena Schwarting and Matthias Rick and Kerstin Schill and Christof B{\"u}skens},
	title = {Autonomous In-Ice Exploration of the {Saturnian} Moon {Enceladus}},
	booktitle = {69th International Astronautical Congress (IAC)},
	year = {2018},
	month = oct,
	organization = {International Astronautical Federation},
	paper = {IAC-18-A3.5.7.x44627},
	abstract={The data collected by the Cassini spacecraft provide strong evidence for the existence of liquid water under the
	ice-covered surface of Saturn's moon Enceladus. For that reason, it is currently one of the most promising candidates to
	find extraterrestrial life in our solar system. The Enceladus Explorer (EnEx) initiative of the German Aerospace Center
	proposes a concept for a lander mission to probe that moon for life, which uses one maneuverable melting probe to
	retrieve a sample from the sub-glacial water reservoir as well as multiple smaller melting probes as navigation aids. In
	addition to technical challenges, a major prerequisite for the success of that mission is the ability to autonomously
	navigate through the ice of Enceladus. We present an integrated system for multi-sensor fusion, high-level planning
	and autonomous decision making, as well as trajectory optimization and optimal control, which aims to solve that
	challenging task. The sensor fusion module is responsible for localizing the probes as well as creating a map of the
	environment. It uses an approach based on graph optimization techniques, which simultaneously estimates the position
	and attitude of all probes in 3-dimensional space. For mapping, an extended formalism for representing uncertainty is
	utilized, which allows one to distinguish between different causes of uncertainty. Based on that information, the autonomy
	module decides on the further course of action. It creates a high-level plan in order to reach the mission goal
	or sub-goals, e.g., improving the localization by taking further measurements. In addition, autonomous fault detection
	is performed, and if required, the system's behavior is adapted in order to compensate for degraded components or
	unexpected situations. For realizing the high-level plan, the guidance module computes a trajectory in consideration
	of different optimality criteria. Finally, a model predictive control algorithm generates and applies the corresponding
	control commands and ensures a reliable tracking of that trajectory. Both trajectory planning and control take infor-
	mation provided by the sensor fusion into account. Furthermore, in order to cope with the high degree of uncertainty
	regarding environmental conditions, the model of the system dynamics is improved by means of automated parameter
	identification and adaption. In addition to a description of our system, we present preliminary results from a field test
	campaign on a European glacier with the entire EnEx system.}

}