Human Brain as Model:

If we observe the structure of a human brain, it is revealed that there is not much space in between the ears, however all the brain matter residing in that small space is capable of functions which even the smartest computers can never do. Interpretation of the world by visual means and understanding the speech also comes fairly easy and with efficient energy than any possible computer system.

Research on Neuromorphic Chips:

Therefore corporate and academic labs have performed experiments with chips which are modeled on the features observed in brains. Such types of chips are known as ‘neuromorphic’ chips. Such chips contain networks of nerve like cells or neurons which communicate by the spike in electricity. These chips are fairly more efficient in energy than the conventional chips. Some of these chips could also reprogram themselves and learn new features and skills.

Drone Intelligence Demonstration:

This chip was tested by installing it on a small drone aircraft which weighs around 100 grams or less. During the experiment, this prototype chip including 576 neurons made of silicon, received data from this drone’s ultrasound, infrared and optical sensors during its flight between 3 separate rooms.

In its flight through each of the rooms, the incoming sensor received a new pattern of data by bouncing from furniture, walls etc. An electrical activity was created in neurons. This in turn resulted in a report that the drone had entered a new room and the neurons reconnected themselves with every new change and learnt as per the new data. This was a way for the chip to mimic a human brain. Therefore, the next time this drone entered into the same space, it recognized and identified it by a signal.

Energy Efficient:

For practical usage, the chip still needs more improvements but the idea has received motivation and support required for the research that is needed in the field of neuromorphic chips.  The craft had been specially designed for this test by the company Aerovironment which makes drones. The dimensions are a square craft sized 6 inches, with a height of 1.5 inches and weighing around 93 grams (with the battery). The chip weighed 18 grams and needed 50 milliwatts power to run. This is very energy efficient, as any conventional computer couldn’t run such intelligent software on this power.

Future of Neuromorphic Chips:

Similar prototypes have provided promising results; however, a lot of work remains to be done before any such technology could be utilized for beneficial work. General Motors and Boeing are already interested in using these chips for creating intelligent sensors for cars and planes.