Connect with us

Featured

MIT’s new mapping technology makes drones wicked fast and great at dodging things

Avatar

Published

on

Scientists at MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) have developed small, ultra-quick drones, capable of autonomously navigating through unfamiliar environments at great speed.

The drones use a system called NanoMap, which models uncertainty of the drone’s position over time to more reliably navigate through new places. It uses depth-sensing technology to gain a sense of its surroundings and can simultaneously consult a database of all the places it has ever travelled to predict how to most effectively move through environments it can’t immediately see.

 

Traditional drones navigate by taking detailed measurements of all objects surrounding it at all times using a system called SLAM (Simultaneous Localization And Mapping). This means they have a great awareness of their environment but they also travel and react slower than drones using MIT’s system. Basically, the idea behind MIT’s innovation is that sometimes, in order to get to where you need to go as quickly as possible, you may have to sacrifice a little bit of precision.

Rather than attempting to map an area perfectly, MIT’s system use a fuzzy logic called modelling uncertainty, designed to generally avoid any objects that a drone flies towards but without giving a complete picture of everything in the vicinity. In their testing to date, these drones have crashed only 2 percent of the time. The scientists say that had they not been using modelling uncertainty, the drones would have crashed 28 percent of the time.

Part 2 on next page.

Prev Page1 of 2
Use your ← → (arrow) keys to browse

Advertisement
Comments

Our Videos