Artificial cerebellum enables human-like object handling by robots
July 10, 2012
Biological circuitry projection of recurrent forward control loops in biologically inspired adaptive microcircuit (credit: N. R. Luque, J. A. Garrido, R. R. Carrillo, S. Tolu, E. Ros/International Journal of Neural Systems)
University of Granada researchers have developed an “artificial cerebellum” that controls a robotic arm with human-like precision.
The cerebellum is the part of the human brain that controls and coordinates body movements.
Existing robots typically require strong forces and are power-consuming. This approach is not compatible with humans because a malfunction might be potentially dangerous.
To solve this, University of Granada researchers have implemented a biologically inspired adaptive microcircuit based on a new cerebellar spiking model that adapts to corrections and stores sensory effects.
It also records motor commands to predict the action or movement to be performed by the robotic arm.
The biologically inspired architectures used in this model combine the error training approach with predictive adaptive control. The robot performs automatic learning and two control systems enable accurate and robust control of the robotic arm during object handling.
The cooperation between the artificial cerebellum and the automatic control system enables the robot to adapt to changing conditions and interact with humans.
N. R. Luque, J. A. Garrido, R. R. Carrillo, S. Tolu, E. Ros, Adaptive Cerebellar Spiking Model embedded in the control loop: Context switching and robustness against noise, International Journal of Neural Systems, 2011, DOI: 10.1142/S0129065711002900
Comments (7)
by asiwel
Research like this is exciting and amazing. Just a few days ago Kurzweil.net reported on work involving human-like robot-walking gaits using a “half-center neural central pattern generator” modeled on the human neural net that controls walking rhythms (First robotic legs to fully model biologically accurate walking, July 6, 2012). That circuit and pair of robot legs was “not connected” to the central nervous system and balancing sensorium. But it looks like this “artificial cerebellum” circuit (which evidently is already being used with artificial arms) would be the ideal thing to add next to provide additional inputs and control signals for the CPG and robot legs system. That would begin to be really impressive.
by josdorpjossie
This looks a little bit like the microcore that Mark Tilden invented many years ago. Admitted, that was a lot more primitive, but I’m a bit surprised this is the very first time someone built something like this. I can’t believe these smoothly walking robots I see on youtube can work without these kind of biological inspired mechanisms..
by MrFriendly
I’m wondering how many neurons are modeled. I have no idea how many cells exist in a real cerebellum.
by Editor
256 Mossy fibers, 1500 Granular layer cells, 48 climbing fibers, 48 Purkinje cells in forward architecture and 96 in recurrent architecture, 24 Deep Cerebellar Nucleus cells in forward architecture and 48 in recurrent architecture.
by Bri
It’s being developed for all robotic applications, and pure science. Most research is for basic understanding, not necessarilly system specific. All application must then compete. Those that are useful will be conserved.
by Gorden Russell
Before this is used for cleaning robots it will be used in factories where robots build robots.
by Barrett Hoffarth
I am curious what activities this system is involved in? It would be interesting to see this placed into an average home and used to clean up. This kind of technology could move far beyond the Roomba if it is designed correctly.