When humans move in their environment, they constantly check the terrain in front of them for obstacles, depressions or moving persons and vehicles. Not walking against the proverbial lantern post or falling into the next storm drain is just one of the reasons for making this behavior so important for us.

For an autonomous or semi-autonomous vehicle, this is also one the most crucial behaviors. The riskiest situations in which an autonomously driving robot can find itself, arise if it cannot reliably detect obstacles in its path.

Fig. 1: Scanning the environment with a laser scanner warns the robot software about dangerous obstacles or depressions.

By using a laser scanner attached to a motor as it is integrated in the CHRYSOR robot, a 3D terrain analysis can be performed (see figure 1). The resulting data (see figure 2) can be used for various purposes:

• Detect obstacles in front of the vehicle
• Measure the size of certain objects.
• Check if an object has disappeared or changed.
• Place the vehicle in a certain distance and orientation to a measured object (e.g. a wall or guard house).
• Verify if a door or gate has been closed.

Fig. 2: 3D environment scan taken with a laser scanner that was rotated by a motor.

When using the laser scanner while driving, a terrain height map can be created with the help of the Robowatch Localization library. The Localization library delivers a precise orientation and position, enabling the software to integrate the individual laser scanner measurements (see top-right of figure 4) to a height map (see figure 3 or the left side of figure 4).

Fig. 3: Terrain height map displayed in our visual tool Visdo.

In some cases, this height map might still be too noisy to detect obstacles correctly. A smart statistical analysis algorithm processes the map data and laser scan time stamps to eliminate noise. Consequently, the software can calculate an obstacle map (see figure 4), depending on the size of the specific vehicle. This obstacle map can be used very effectively to avoid obstacles or follow an off-road path.

Fig. 4: Smart statistical analysis algorithms allow transforming a noisy height map (left) into an obstacle map (right). Obstacles are displayed in pink. Blue is a safety perimeter around the obstacles which the center of the vehicle should not cross. An individual laser scanner measurement is displayed in red on the top-right.

Our engineers will customize the appropriate terrain analysis solutions for your application. We offer consulting services to support your company in building your own military or industrial applications of terrain analysis and integrating them into a larger system. Our terrain analysis algorithms are a powerful tool, because they allow the detection of obstacles, the free path in front of a vehicle, environment changes and more.