Neuro-electronic hybrid systems present an opportunity to harness the computational and learning capabilities of neuronal circuits that allow them to solve many types of problems. We use neuronal cultures grown on multi electrode arrays to explore methods to encode inputs to the culture, evaluate decoding schemes for activity seen and observe the changes happening over short and long periods.
We have been able to use temporally coded voltage pulses to stimulate the cultures and use an array of perceptrons to decode the activity. This can be used within the framework of liquid state machines for solving problems.
We have implemented a real-time closed loop system using the above scheme to control a robot exploring an arena while avoiding the obstacles. We aim to use such platforms to explore features of neuro-electronic hybrid systems.
|
Different patterns of activity evoked from the culture on using temporally coded inputs. |
Schematic of the neuro-electronic hybrid system. |
Components of the neuro-electronic hybrid system. |
 Professor |
 Professor |
 Professor |
