Robot with LiDAR Laser Explores Danger Zones

Robot systems explore unfamiliar terrain, buildings or danger zones with cameras

In a disaster such as a chemical plant incident or flooding, emergency services need ways to quickly get an overview of the situation. But in many cases, they are not permitted to enter the scene itself in order to avoid putting themselves at risk.

Researchers at Fraunhofer FKIE in Wachtberg are working on a solution to this problem in the 3D-InAus project. A robot equipped with a LiDAR (light detection and ranging) laser explores the area. LiDAR technology uses pulses of light to scan the environment and measure distances. This creates a 3D model showing buildings, rooms, open space, objects and all of the associated dimensions and distances. Users can move freely around the 360-degree visualization using a joystick as they investigate the virtual environment.

Timo Röhling, a technical project manager from the Cognitive Mobile Systems department, explains: “Compared to robot systems that use cameras to explore a danger zone, our project goes a big step farther. The laser pulses supply measurements for precision 3D cartography of an area of terrain or building. Distances and dimensions are not estimated but instead determined with accuracy down to just a few centimeters.”

Geometric point cloud derived from laser pulses

The centerpiece of the hardware is a LiDAR laser mounted on a turntable. A rotating mirror is built into the LiDAR module. It can scan an area made up of 16 vertical sections, or “slices,” ten times per second. The turntable rotates the laser so that the vertical sections cover an entire 360-degree view along the horizontal axis. The system generates a total of 1.3 million laser pulses per second. These pulses bounce off the surrounding objects, and the time lag in between is used to calculate the relevant distance. The LiDAR module is mounted on a vehicle that moves around through the area, either continuously or in stop-and-go mode. Continuous operation is much faster but also less accurate. The result is a 3D point cloud in which each point stands for a laser pulse or distance measurement. There is also a camera system with up to six cameras. The images from the cameras are used to color the associated objects or shapes. “You might think of us melding the camera images and point cloud together. This gives us a vivid, detailed and also geometrically accurate 3D environment showing buildings, open space and objects,” Röhling says.

The raw data acquired by the LiDAR laser is pre-processed by a computer module inside the robot before it even finishes its mission. The final visualization is then produced on a stationary basis during post-processing. It takes about three hours to map an area measuring 400 x 400 m. In a disaster when time is of the essence, accelerated operation can be used to get an initial overview in as little as one hour. It is also possible to explore an area using multiple vehicles simultaneously.

More reading here