MIT robot uses rangefinder, screams of bystanders to navigate high-speed flight

3-D map, rangefinder/locator allows micro-UAV to navigate through parking garage, very quickly

Here’s a great idea: If you don't like the slow, careful programming that makes self-driving vehicles and small robot aircraft move like elderly humans who stop every few seconds to remember which direction they were walking, just speed those puppies up until you terrify everyone in the building and drive the insurance rates up.

That's what some out-of-control UAV cowboys did at MIT's Robust Robotics Group, by installing a 3-D map in its memory, an electronic device called an Inertial Measurement Unit (IMU) that measures velocity, orientation and the effect of gravity to estimate its position and a laser range-finder to estimate the distance and time-to-impact of the nearest obstacle.

The robust roboticists added a gloss of respectability to their suicide flier (which might actually be a homicide flier, since it appears able to avoid destroying itself on every flight), by presenting a paper weighed down with a title very nearly dull enough to conceal their true intentions (State Estimation for Aggressive Flight in GPS-Denied Environments using Onboard Sensing {Zzzz}).

The key to the setup is the detailed 3-D map in the flier's memory, according to IEEE Spectrum. Following a map that already exists is easier than the alternative – simultaneous location and mapping (SLAM), but much harder than avoiding collisions by noting the location of beacons in key positions and using those to identify position, which Spectrum refers to as "sort of cheating…because you'll never get it to work outside your lab."

The 3-D map provides an excessive but acceptable level of help to the UAV, which would need more computing power (more weight) and move more slowly if it had to recognize obstacles on its own and try to avoid them (by flying slowly).

Truly autonomous UAVs are still cutting edge in the robotics world; successful long-range robotic UAVs such as the Air Force Predator killer drone are remote controlled, which is an even bigger cheat than using location beacons.

The MIT crew went even further into ambitious autonomy in 2010 with a version of the same killer toy responding to voice commands that were made possible due to research that created "models of spatial prepositions and motion verbs," giving the craft a better understanding of the desires of others than most people who attend MIT.

Other roboticists are also working on "aggressive maneuvering" involving flying through windows and hula hoops, in an attempt to build robots able to fly with the speed and dexterity of "birds flying through urban environments," according to the Robust Robotic paper written by Adam Bry, Abraham Bachrach and Nicholas Roy of MIT.

So far no other aggressive experiments has been able to move as fast or be as randomly threatening to passersby as MIT's most recent experiment buzzing a parking garage using a "fixed-wing micro-air vehicle" (tiny remote-control airplane without the remote control).

Watching the video gives one hope for the future.

Deciding never to go near MIT or that parking garage gives one even more.

Read more of Kevin Fogarty's CoreIT blog and follow the latest IT news at ITworld. Follow Kevin on Twitter at @KevinFogarty. For the latest IT news, analysis and how-tos, follow ITworld on Twitter and Facebook.

From CIO: 8 Free Online Courses to Grow Your Tech Skills
Join the discussion
Be the first to comment on this article. Our Commenting Policies