New drone from MIT is never sure where he is — and it helps him to avoid collisions

Oddly enough, but the management system of modern drones “suffer” from excessive precision. Often require hundreds of measurements to the UAV knew exactly in which point of space he is. In case there are any inaccuracies, he simply “flies” to the nearest obstacle. However, to load such a volume of information in a small drone is not always possible.

New mapping system NanoMap, developed in the laboratory of computer Sciences and artificial intelligence (CSAIL) Massachusetts Institute of technology (MIT), provides a harmonious balance between accuracy and speed. The drones can travel at speeds up to 32 km/h in places that are “inhabited” by lots of items, for example, in a dense forest or in the city.

In contrast to other mapping systems (e.g., SLAM) algorithm NanoMap the principle of uncertainty during movement. The system measures the depth of the surrounding space, and then separate “pictures” forms a single 3D image. Now the drone can more accurately assess the environment and predict its flight relative to the encountered objects.

If without a system of NanoMap every fourth test flight of the drone ended with a collision, setting the number of collisions decreased to 2 %.

Source — Engadget

Leave a comment

Confirm that you are not a bot - select a man with raised hand: