3D Local Multiresolution Grid for Aggregating Omnidirectional Laser Measurements on a Micro Aerial Vehicle

Konferenz: ISR/Robotik 2014 - 45th International Symposium on Robotics; 8th German Conference on Robotics
02.06.2014 - 03.06.2014 in München, Germany

Tagungsband: ISR/Robotik 2014

Seiten: 7Sprache: EnglischTyp: PDF

Persönliche VDE-Mitglieder erhalten auf diesen Artikel 10% Rabatt

Autoren:
Droeschel, David; Behnke, Sven (Autonomous Intelligent Systems Group, University of Bonn, Germany)

Inhalt:
Micro aerial vehicles (MAV) pose a challenge in designing sensory systems and algorithms due to their size and weight constraints and limited computing power. We present an efficient 3D multiresolution map that we use to aggregate measurements from our lightweight continuously rotating laser scanner. We estimate robot motion by means of visual odometry and scan registration, aligning consecutive 3D scans with an incrementally built map. Hence, we are able to aggregate scans over time and increase the density of the map. By using local multiresolution, we gain computational efficiency by having a high resolution in the near vicinity of the robot and a lower resolution with increasing distance from the robot, which correlates with sensor characteristics in relative distance accuracy and measurement density. Compared to uniform grid-based maps, multiresolution leads to the use of fewer grid cells without loosing information and consequently results in lower computational costs. Hence, we are able to efficiently register new 3D scans with the map, utilizing the multiresolution property in a coarse-to-fine approach. In experiments, we compare the laser-based motion estimate with ground-truth from a motion capture system as well as a state-of-the-art registration method. Overall, our approach allows for building accurate 3D obstacle maps and estimating the MAV trajectory in real-time by 3D scan registration.