There’s a lot of talk about our drone-powered future, certainly for deliveries and other tasks, but to get to that reality, drones need to be an awful lot smarter than the ones you mostly see flying around at the moment.
For example, drones used for commercial activity would need to be able to communicate with each other effectively, and on top of that, they need to be able to reassure the populus that they’re safe, that drones flying around making deliveries should be an acceptable vision of the future.
Part of that safety is the ability to autonomously avoid obstacles in areas that they aren’t familiar with – a drone avoiding obstacles that it already knows the position of isn’t really replicating real-life.
To achieve this, Andrew Barry, an MIT Computer Science and Artificial Intelligence Lab (CSAIL) student, created a stereo-vision algorithm that lets the drone create a (frankly cool) map of its surroundings in real-time, all using the on-board processing power of the drone.
The drone, which uses a traditional aeroplane-like design rather than a quad-copter, has a camera on each wing and two processors “no fancier than you’d find on a cellphone.”
Read next: What analyzing 1 million tweets taught us