Making 3D Maps on the Move
A vehicle uses off-the-shelf components to build 3D maps of an area.
At a robotics conference last week, a vehicle called ROAMS demonstrated a cheap approach to mobile map-making.
ROAMS (Remotely Operated and Autonomous Mapping System) was created by researchers at the Stevens Institute of Technology in Hoboken, NJ, with funding from the U.S. Army. It uses several existing mapping technologies to build 3D color maps of its surroundings, and it was demonstrated at the 2009 IEEE conference on Technologies for Practical Robot Applications in Woburn, MA.
The system uses LIDAR (Light Detection and Ranging), which involves bouncing a laser off a rapidly rotating mirror and measuring how the light bounces back from surrounding surfaces and objects. The same technology is already used to guide autonomous vehicles, to make aerial maps, and in spacecraft landing systems.
A conventional 3D LIDAR system, which consists of several lasers pointing in different directions, costs over $100,000. The Stevens researchers created a cheaper mapping system by mounting a commercial 2D LIDAR sensor, which costs about $6,000, on a pivoting, rotating framework atop the vehicle. While the system has a lower resolution than a regular 3D LIDAR, it could still be used for low-cost architectural surveying and map making in military situations, the researchers say. “The prototype system is around $15,000 to $20,000,” says Biruk Gebre, a research engineer at Stevens who demonstrated the device.
The system takes about 30 seconds to scan a 160-meter-wide area. A color camera also on the rotating frame provides color information that is added to the map later on. And the Stevens researchers developed a way to maintain the same resolution by automatically adjusting the scanning process depending on the proximity of objects. A human operator rides in a larger vehicle that follows the robotic one from up to a mile away, says Kishore Pochiraju, professor and the director of the Design and Manufacturing Institute at Stevens. Ultimately, says Pochiraju, “we want to leave this robot in a location and ask it to generate a complete map.” Such a vehicle could, for example, drive into a dangerous area and generate a detailed map for military personnel.
“They’re using a relatively low-cost system,” says John Spletzer, an associate professor at Lehigh University who uses similar technology to create autonomous wheelchairs. “There’s a lot of groups working on it; it’s pretty interesting.”
Nicholas Roy, an associate professor at MIT who develops autonomous and self-navigating vehicles, also notes that other research groups have developed similar technology. He says that the biggest challenges in autonomous map-making are identifying obstacles and sharing mapping between several robots.
The AI revolution is here. Will you lead or follow?
Join us at EmTech Digital 2019.