From Neuro-Imaging to Neuro-Robotics
With 50 million neurons (processing elements) and several hundred kilometers of axons (wires) terminating in almost one trillion synapses (connections) for every (!) cubic centimeter, and consuming only about 12 watts energy for the entire cortex, the brain is arguably one of the most complex and densely packed, yet highly efficient information processing systems known. It is also the seat of sensory perception, motor coordination, memory, and creativity – in short, what makes us humans to humans. There are fascinating complementarities between human brains and artificial information processing systems created by humans: what appears to be challenging for us, seems easy for computers, while highly demanding problems for computers and robots such as face recognition and smooth locomotion are mastered without much difficulties by humans. I will argue in this talk that many of the computationally hard problems of perception and action appear to be easy to humans precisely because they have been successfully solved by the brain during the course of its evolution and ontogeny. The corollary of this claim is that the structure and function of the brain may already contain solutions to many of the hard problems faced by artificial intelligence and cognitive robotics systems today. To this end, I will first review some of the recent neuro-imaging studies elucidating the functional and connectivity architecture of the mammalian brains. I will then try to convince you that time is ripe for communication engineers, circuit designers, and roboticists to join with brain scientists to start reverse engineering the whole brain, which in turn will lay ground for new generations of truly disruptive neuroengineering applications.