Publication
Probabilistic Combination of Heuristic Behaviors for Shared Assistive Robot Control
Felix Goldau; Udo Frese
In: PETRA '24: Proceedings of the 17th International Conference on PErvasive Technologies Related to Assistive Environments. International Conference on Pervasive Technologies Related to Assistive Environments (PETRA-2024), June 26-28, Crete, Greece, ISBN 979-8-4007-1760-4, ACM, 2024.
Abstract
Technology in general is developed to improve the lives of their users, often by allowing them to handle individual struggles. Assistive robotics takes this concept to its extreme by (re-) enabling users to physically interact with their environment in their daily life. For a successful utilization however, a user interface is required that allows for easy and quick interaction. Based on the promising concept of Adaptive Degree of Freedom (DoF) Control, this paper presents a novel heuristic implementation of the underlying principle, without relying on learning user actions.
To achieve this, a mixture distribution is obtained which expresses how likely the user wants which motion. Here, every mode of the mixture represents a heuristic behavior. Each such behavior defines its own distribution of motion, as well as a weight indicating how likely it is in the current situation. The best fitting DoF is obtained from this mixture and offered to the user with an interface.
This general-purpose control method has been tested in a small technical study, the results of which show its general viability, promising chances for a significant reduction of mode changes, as well as very good quantitative feedback b