|Institution||Claude Bernard University of Lyon 1|
|Location||Nautibus (Université Lyon1)|
|antoine.grea at univ-lyon1.fr|
|Subject||The subject of this thesis focuses on the design of intelligent systems for robots to support dependents. This intelligence must allow the robot to gain more autonomy in its decision.|
|Abstract||The interaction human-robot is a complex and varied theme. As a human, we are demanding our social interactions, our interlocutors we must understand and interpret our requests.
The robots often need to have specific orders in advance to operate. Here it is question of solving this problem with an intelligent system for recognizing intentions. Depending on observed user actions, it should be possible to derive a plan of action and a predictable order to know the best way to help it. For this we use a reverse planning system that is able to know the most logical plan that integrates the observed actions. This plan is recalculated for each observation. That's why we use a new plan repair method to fix only new inconsistencies brought by observations in the system.
We hope to create an accurate and robust system that will help the user when they need it without the need for explicit interactions with the robot.
Last update : 2017-03-24 09:45:05