Last year, a report from research group IDTechEx predicted that delivery drones will not become mainstream for at least 10 years because of prominent concerns including the risk of accidents.
The world is an uncertain place for drones delivering packages through populated cities or navigating through the many trees and obstacles in the forest.
"Overly confident maps won't help you if you want drones that can operate at higher speeds in human environments", said computer scientist Pete Florence, lead author on a related paper. "An approach that is better aware of uncertainty gets us a much higher level of reliability in terms of being able to fly in close quarters and avoid obstacles".
Unlike other common mapping systems, such as simultaneous localization and mapping (SLAM), which are data intensive and hard to maintain at real-time, the NanoMap uses depth-sensing to measure just the drone's immediate surroundings.
For example, if NanoMap was not modelling uncertainty and the drone drifted just five per cent away from where it was expected to be, the drone would crash a few times every four flights.
Without the element of uncertainty, if the device drifted just 5 percent away from where it was expected to be, it would crash a couple of times per four flights.When accounting for uncertainty, that rate is reduced to 2 percent. When it accounted for uncertainty, the crash rate dropped to one in 50 flights.
A paper co-written by Florence and MIT professor Russ Tedrake, with research software engineers John Carter and Jake Ware, is available to read online. It was recently accepted to the IEEE International Conference on Robotics and Automation (ICRA), which takes place in May in Brisbane, Australia.
The output of SLAM methods are not typically used to plan motions. To compensate, researchers often use methods like "occupancy grids", where many measurements are incorporated into one specific representation of the 3D world. At high speeds, computer-vision algorithms can not make much of their surroundings, forcing drones to rely on inexact data from the inertial measurement unit (IMU) sensor, which measures things like the drone's acceleration and rate of rotation. Simply put, the system doesn't sweat the details.
MIT CSAIL's system enables drones to fly through forests (Photo credit: Jonathan How, MIT).
"It operates under the assumption that, to avoid an obstacle, you don't have to take 100 different measurements and find the average to figure out its exact location in space", according to MIT.
The key difference from previous work, said systems scientist Sebastian Scherer from Carnegie Mellon University's Robotics Institute in Pennsylvania, is that the system creates maps of images with "position uncertainty" built in rather than just sets of images with their positions and orientations. "Keeping track of the uncertainty has the advantage of allowing the use of previous images even if the robot doesn't know exactly where it is and allows in improved planning".
NanoMap is especially useful among small drones traveling through small spaces, for the objective of search-and-rescue, defense, package delivery, and entertainment.
Other teams are also tackling the issue of drone navigation in tight areas.
This work was supported in part by DARPA's Fast Lightweight Autonomy program.