Researchers funded by the Defense Advanced Research Projects Agency (DARPA) have tested a drone that can learn to do things on its own.
This capability is achieved thanks to a prototype of a neuromorphic chip that mimics human brain functions, making it a brain chip, according to Gizmodo. The research was completed by a group of scientists at HRL Laboratories' Center for Neural and Emergent Systems.
The drone consists of 576 silicon neurons designed to respond to data collected from optical, ultrasound, and infrared sensors. These neurons use spikes of electricity to communicate with one another.
The prototype for the drone and the chip are part of DARPA's Systems of Neuromorphic Adaptive Plastic Scalable Electronics (SyNAPSE) project that is focusing on giving robots the same biological brain activity possessed by humans, RT reported. HRL Laboratories is working with IBM Research and Hewlett-Packett on the project, having received $34.5 million in funding from DARPA six years ago, while IBM received $42 million.
The 18-gram chip uses 50 milliwatts of power. Silicon Valley-based Aerovironment is responsible for creating the drone prototype, designing it to weight 93 grams, including the battery, and measure six inches square.
The team's machine was shown in the MIT Technology Review to be flying in three different rooms, where sensor data collected from the walls, furniture and other objects produced a pattern of electrical electricity "that the chip has never experienced before," which helped the drone understand the layout of the rooms and how it should go about its way when in them, RT reported.
"This showed it is possible to do learning literally on the fly, while under very strict size, weight and power constraints," said Narayan Srinivasa, head of HRL's Center for Neural and Emergent Systems.
Srinivasa added that commercialization of these robots is already a goal for major tech companies, such as Boeing and General Motors. The race to develop this technology shows that super intelligent drones may be more common in the near future.
