That is exactly what a team of researchers built into their vehicles, however, as they aimed to create a system for consistent, safe drone flight through real-world environments.
The group, from the Massachusetts Institute of Technology’s (MIT) Computer Science and Artificial Intelligence Laboratory, said their NanoMap system could be useful for everything from search-and-rescue to military or delivery applications, in tight urban environments such as warehouses or natural areas like forests.
Many existing autonomous movement systems rely on intense on-board computing and intricate pre-made maps, which might not reflect changes in the real world. High speeds are also challenging for computer vision algorithms, forcing drones to rely on inexact measurements of acceleration and rotation to navigate.
Instead, the MIT system considers the drone’s position over time to be uncertain, and accounts and models for that uncertainty. Instead of working out exactly where an object is with 100 different measurements, a depth-sensing system gathers enough information to know that an object is in a general area.
“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.”
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.
In tests, if NanoMap did not model uncertainty and the drone drifted just 5% from its expected position, it crashed more than once every four flights. When it accounted for uncertainty, the crash rate dropped to one in 50 flights.
The cluttered environment
Other teams are also tackling the issue of drone navigation in tight areas. At the University of the West of England, researchers at the Unmanned Flight Laboratory used a more computing-heavy approach with their Agile Drone Three project.
“The parameters that we are working to are slightly different in that we have a clearer view, the cluttered environment is more distinct,” said avionics engineer and associate professor Steve Wright to Professional Engineering. “We have got a low level of uncertainty and we do that with having more sensors.”
The higher level of certainty could allow much faster flight, he added, with potential speeds of up to two or three times as fast as the MIT group's drone.
Another potential issue with the NanoMap system is how likely regulators are to allow uncertainty in autonomous systems. 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 MIT paper, which was supported by the US defence research agency DARPA, was recently accepted to the IEEE International Conference on Robotics and Automation, which takes place in May in Brisbane, Australia.
Content published by Professional Engineering does not necessarily represent the views of the Institution of Mechanical Engineers.