MAVERic is a versatile robotic vision head developed for oculo-motor
research at the RIKEN
Brain Science Institute's
Laboratory for Mathematical Neuroscience,
in collaboration with the University of Southern
California's Computational Learning and
Motor Control Lab. The hardware was built by SARCOS.
It has 7 Degrees of Freedom (DOF) - pan and tilt in each eye and head
rotate, nod and tilt - and is controlled using a real time operating system (vxWorks).
MAVERic is equipped with multiple sensory modalities including position sensing (7DOFs),
load sensing (3DOFs), dual microphones,foveal and peripheral vision in each eye and
a 6-axis gyroscope in addition to laser range finders in each eye.
The main emphasis of the project revolves around exploring principled frameworks for mulitmodal sensor and cue integration while incorporating the constraints imposed by the complete sensory-motor loop. Example projects include real-time auditory source localization, online reinforcement learning using stochastic gradient techniques for task learning, real-time learning of oculomotor behaviors (VOR, OKR, smooth pursuit), multi-modal visual attention, cue/state prediction and inference based on Particle Filters etc.
The hardware components of this head prototype form a realistic model of a human head, hence, lending it ideal as a testbed for the development of online vision, audition, and learning algorithms.