SFF-based aerial vehicle to the rescue

A team of students from MIT won the fifth mission of the International Aerial Robotics Competition (IARC) held by the Association for Unmanned Vehicle Systems International (AUVSI). For their performance, the MIT students were awarded a $10,000 prize and received their $1,000 entry fee back as an incentive to encourage teams to put forth their best effort during the first year of a mission. This is the first time in 19 years of the competition that any team has won during the first year of a mission.

The IARC, which started in 1991 at the Georgia Institute of Technology, is the longest-running robotics competition in the world. Teams develop Micro Aerial Vehicles (MAVs) to perform missions never before required by any flying object. AUVSI presents competitions for fully autonomous vehicles on land, sea, and in the air.

Unmanned Aerial Vehicles (UAVs) and MAVs rely on the availability of global positioning infrastructure including satellite-based GPS and on developing algorithms to pinpoint their position. However, most indoor environments and many city street canyons surrounded by steep walls on both sides of the street are not accessible to external positioning systems, limiting the ability for autonomous MAVs to fly through these areas.

To meet the challenges of this competition, the MIT team used the Pelican model quadrocopter (Figure 1) from Ascending Technologies, Germany, with a laser scanner that creates a map of its surroundings and an optical system that aids in the vehicle’s relative motion. The team integrated a business-card-sized module (58 mm x 65 mm) from LiPPERT Embedded Computers, Germany, loaded with an Ubuntu version of the Linux operating system to run the algorithms in real time.

21
Figure 1: The Pelican quadrocopter from Ascending Technologies uses a business-card-sized module loaded with Ubuntu Linux to run algorithms in real time.
(Click graphic to zoom by 1.9x)

Equipped with LiPPERT’s CoreExpress-ECO module (pictured on page 11), the four-propeller quadrocopter MAV is a light (28 g) and small autonomously flying netbook. It uses an Intel Atom Z530 processor on the legacy-free CoreExpress module. CoreExpress is an open standard (downloadable at www.sff-sig.org/coreexpress.html) that accommodates the latest version of PCI Express and USB 3.0. An ARM-based processor directed by the CoreExpress board controls the quadrocopter’s motion.

The MIT students equipped their MAV with a laser range finder to estimate position, yaw angle, and altitude within a 4 m range in a 240° angular field of view. The team’s MAV uses an algorithm based on the Belief Roadmap (BRM) algorithm, which is a generalization of the Probabilistic Roadmap (PRM) algorithm using an Extended Kalman Filter (EKF) to find the minimum expected path. In this competition the MAV had to fly in an unknown building, explore and map it (detecting blind alleys), find a hidden object, take pictures, radio them back to the base, and return to base autonomously without GPS or a human pilot while avoiding collisions with objects, walls, ceilings, or floors.

During the IARC’s third mission in 2000, an aerial robot from the Technische Universität Berlin (TU-Berlin) detected and avoided obstacles, many of which could have destroyed the robot; identified which people were dead and alive in a simulation, distinguishing them based on movement; and relayed pictures of the survivors along with their locations back to rescue personnel. This mission was specified by the U.S. Department of Energy’s Hazardous Material Management and Emergency Response (HAMMER). The sixth mission, which has already been specified by the IARC, will demand more advanced behaviors than are currently possible in any existing aerial robot.

For more information, contact Hermann at hstrass@opensystemsmedia.com.