A recent DARPA project that piggybacks a “neuromorphic chip” atop a minidrone shows promise in developing computer chips that may someday be able to learn much in the way the human does.
UAV manufacturer Aerovironment provided a small drone (six-inches square, 1.5 inches high and weighing 93 grams) to HRL Laboratories’ Center for Neural and Emergent Systems to act as a host for a chip that is designed to operate like a biological brain. With 576 silicon neurons, the drone can synthesize the data from its perceptual sensors via the neuromorphic chip and reportedly performed well in experiments.
MIT Technology Review reports that the drone – after flying through three rooms — exhibited behavior that resembles learning:
“The first time the drone was flown into each room, the unique pattern of incoming sensor data from the walls, furniture, and other objects caused a pattern of electrical activity in the neurons that the chip had never experienced before. That triggered it to report that it was in a new space, and also caused the ways its neurons connected to one another to change, in a crude mimic of learning in a real brain. Those changes meant that next time the craft entered the same room, it recognized it and signaled as such.”
The experiment is part of a DARPA challenge to “develop low-power electronic neuromorphic computers that scale to biological levels.”
Recognizing that computer systems are limited by power constraints when processing data, the agency hopes to develop AI that can mimic our own biological neural functions, which are incredibly energy efficient.
Drones that can learn from mistakes or from abrupt changes in environment could prove quite useful in military applications where battlefield conditions can change in nanoseconds. Drones capable of behavior analogous to human thought could also complete military objectives autonomously should an enemy force cut radio contact between the UAV and its human handler.
Drones that learn could also teach other drones in an autonomous swarming network not to mention the technology’s application in self-driving vehicles.
For consumer users, a learning drone could stop beginners from crashing their drones on often-crash-prone maiden voyages.
Like other nascent experiments, however, researchers really can’t say how the quest to perfect neuromorphic chips in a drone system may benefit society – the possibilities could be endless and far-reaching.
As MIT points out:
“Researchers also face a chicken-and-egg scenario, with chips being developed without much idea of what algorithms they will run and algorithms being written without a firm idea of what chip designs will become established. At the same time, neuroscientists are still discovering new things about how networks of real brain cells work on information. ‘There’s a lot of work to be done collectively between circuit and algorithm experts and the neuroscience community.’”
Jason is a longstanding contributor to DroneLife with an avid interest in all things tech. He focuses on anti-drone technologies and the public safety sector; police, fire, and search and rescue.
Beginning his career as a journalist in 1996, Jason has since written and edited thousands of engaging news articles, blog posts, press releases and online content.
Subscribe to DroneLife here.