Hearings Embodied Artificial Intelligence and Machine Learning

Wednesday, 17.10.2018 - Room 3W03

Esterle Lukas


Title: "Embodied intelligence for autonomous team affiliation in collective robots"

When a collective of robots has to achieve multiple complementary goals in parallel, forming teams to pursue each individual goal becomes essential. In a distributed and highly dynamic system, where goals and resources are not a priori known, a global coordination might not be feasible. In such cases, we have to rely on embodied intelligence, enabling each robot to affiliate itself with a team at runtime. In this talk, we look into the multi-task k-assignment problem where tasks arise dynamically at runtime and require multiple devices to be completed. In these situations, operators might not be able to determine all dynamic aspects of the given environment at the time of deployment. Therefore, we enable each individual robot to reason about its own situation and the state of its own environment when affiliating with a team. To achieve this, we use a novel, bio-inspired approach as well as established machine-learning techniques operating only on aggregated information from the network.


Nach oben scrollen