Sensor fusion for automatic control of  a Bestic robot

Abstract: Designing robots capable of assisting elderly or people with disabilities in their daily life is a challenging and wide task. Many steps of the daily life require assistance as your autonomy diminishes. One of them is eating and that is why for a long time, there has been researches on robots assisting people during their meals. This paper will focus on a BESTIC robot. A BESTIC robot is a device allowing its user to eat meals without external assistance. The purpose of this thesis was to automatize it to make it easier to use for disabled people. This is a delicate process as it requires the robot to be able to perceive food, locate it precisely and take a decision as to which one to pick. The thesis will first focus on the different sensors available for this task and choose the ones the most fit for this task. Will then be discussed the methods available to detect food and the difficulties met during this task. Finally, the thesis will discuss the other problems, especially when it comes to food recognition, and suggestions found for this device. It appears throughout this thesis that the current structure of the robot is not adapted to such a task and that some changes are necessary. It will also appear food detection and recognition are difficult steps that cannot be performed without complex mechanisms in the robot like learning or without advanced sensors to perceive the food.