Robotics and more particularly humanoid robots are envisioned as a solution to support humans in dangerous, repetitive and demeaning tasks. Indeed, humanoid robots, with their human shape, are perfectly tailored to integrate our environment and use tools originally designed for humans. However, unstabilized vision of such robots severely degrades their perception, thus preventing them to autonomously operate in unknown environments. In this context, gaze stabilization emerged as a promising way to overcome this limitation. It consists in actively controlling the motors of the robot head and eyes in order to stabilize the perceived images. In this thesis, driven by bio-inspiration, we explore the potential use of the concept of internal model to support humanoid robot gaze stabilization. This concept from neuroscience, internally simulating the sensorimotor system, is indeed known to play a central role in human motor control and could also benefit to robot control. This doctoral dissertation starts by describing the tools developed to implement and test gaze stabilization controllers. More precisely, dynamic modelling of humanoid robots and middleware based software architecture are addressed. After that, two beneficial uses of internal model for robotic gaze stabilization are demonstrated. First, an anticipatory gaze stabilization based on the concept of virtual linkage is proposed. Then, a multimodal control scheme based on the reafference principle is presented. It complements the first controller with visual and inertial reflexes. ; (FSA - Sciences de l'ingénieur) -- UCL, 2017
Gaze stabilization of humanoid robots based on internal model
2017-01-01
Hochschulschrift
Elektronische Ressource
Englisch
DDC: | 629 |
Adaptive gaze stabilization through cerebellar internal models in a humanoid robot
BASE | 2016
|Multimodal gaze stabilization of a humanoid robot based on reafferences
BASE | 2017
|