You are here

Teaching a plate of brain cells to control a robot

A group of researchers at IISc have managed to "teach" the brain cells taken from a rat and cultured on a glass plate, to help navigate a robot through an arena—while avoiding obstacles.

The researchers took the brain cells of a rat, and allowed them to grow on a specialized tiny glass plate covered with multiple electrodes. They flooded it with a special liquid medium to keep it alive. In a few weeks, the cells grow specialised structures called dendrites, which connect to other cells, thus forming a network. This network starts showing spontaneous electrical activity with generation and transmission of tiny voltage spikes – much like within the brain.

“Our apparatus is basically a mesh of electrical wires embedded on a glass plate that can detect the electrical signals produced by the network of brain cells atop it”, explained Jude Baby George, a graduate student at the Centre for Nano Science and Engineering (CeNSE).

The plate is mounted on an electronics platform that can not only receive signals from the cells, but also send voltage spikes to the tissue via the embedded electrodes. The setup is placed inside a sterile incubator to allow a contaminant free and healthy environment for the tissue to grow. Using a computer and a coding system, the researchers then decode the spiking activity of the network and in turn train the network of brain cells to generate specific signal responses to specific codes.

In one experiment, a robot with infrared light sensors was placed outside in a circular arena. The robot uses the light sensors to "sense" the presence of obstacles in its vicinity and relays it to the electronics platform via a wireless connection. With appropriate programming, the researchers converted the received sensory information from the robots to a specialized stimulation sequence to the brain cells in the glass plate. The brain cells respond to this stimulus by generating their own electrical activity. The software then analyzes this activity and maps it to an appropriate command for the robot so that it can avoid obstacles.

The project began in December 2012 and involved setting up of a sophisticated bio-electronics facility to extract and grow brain cells. Getting the brain cells to grow and thrive required careful control of the laboratory conditions, to ensure no infections affected the cells. And then, once they connected the robot, the abstract code immediately came to life, the result of a successfully “well-taught” bunch of brain cells that could deliver precise electrical instructions to control the robot.

The field of study, known as “Neuro-electronic hybrid systems”, is gaining immense popularity. It involves exploiting the natural learning and processing ability of brain cells to solve problems in an electronic setting. “The learning ability, problem solving skills and adaptability of the human brain have always captivated the interest of researchers”, says Jude. “Such experiments and studies take us closer to understanding how the neuronal system works”. And, potentially build computing systems with “wetware” working synergistically with hardware and software.

 

About the study

The study was presented in the 28th International Conference on VLSI Design and Embedded Systems (2015), for which it was awarded the A. K. Choudhury Best Paper Award.

http://chips.ece.iisc.ernet.in/images/a/ab/Jude_robot_Navigation_VLSI_2015.pdf

http://ieeexplore.ieee.org/xpl/articleDetails.jsp?arnumber=7031714&pageNumber%3D34169%26rowsPerPage%3D100

About the authors

Jude Baby George is a Doctoral student and Grace Matthew Abraham is a project staff member at the Centre for Nano Science and Engineering (CeNSE), Indian Institute of Science.

Bharadwaj Amrutur is an Associate Professor at the Department of Electrical Communications Engineering (ECE) and the Centre for Nano Science and Engineering (CeNSE), IISc. Sujit Kumar Sikdar is a Professor at the Molecular Biophysics unit (MBU) and Center for Nano Science and Engineering, IISc.