Long-term autonomous robotic missions, especially in isolated environments, have to cope with various challenges, e.g., limited possibilities for maintenance or recovery in case of failure. Better adaptation of the robot to its environment is therefore the first step towards a more robust behavioural strategy, leading to less error-prone expeditions and thus reducing the need for potential external maintenance, allowing for a higher degree of autonomy. We propose the integration of episodic long-term memory (ELTM) to existing navigation systems as a building block for such a learning robot system, e.g., improving the efficiency of the robot by avoiding unnecessary replanning. Further, the aim of this work is the investigation of possible benefits of the ELTM regarding different performance metrics, i.e., duration of traverses, computational cost, and improvements in failure resistance.
Episodic Lang-Term Memory for Planetary Robotics Autonomous Navigation
In der Regel sind die Vorträge Teil von Lehrveranstaltungsreihen der Universität Bremen und nicht frei zugänglich. Bei Interesse wird um Rücksprache mit dem Sekretariat unter sek-ric(at)dfki.de gebeten.