When we have time for a vacation, we are often excited to visit the extensive museums in one of the world’s interesting cities.  While walking around the sprawling halls of such a museum, we lose track of how many rooms we have passed and how many corners we have turned. It is difficult for us to guess our position or build a mental map of the museum in our heads. When we suddenly come back to the entrance we notice that we walked a loop. Now everything becomes much clearer. We finally know our position, and this also clears up our mental map of the museum. When we walk that loop for a second time to show our family member the imposing statue that we found in one of the rooms, we feel much surer about where we are and our mental map of the museum takes form. We have essentially done exactly what an autonomous robot will do when orienting itself in a new environment and trying to build a map of it.

The robot needs to know where it is to form a map, but also needs a map to know where it is. Just like you walking in that extensive museum, the robot will do both things simultaneously, improving its mental map while becoming more certain about its position and vice versa. For this reason, the artificial intelligence solution to this problem is called Simultaneous Localization And Mapping (SLAM).

Our software libraries include state-of-the-art algorithms for solving this task. And we still constantly work to make them even better. Based on our SLAM solution several opportunities in relation to your application offer themselves:

• Enable autonomous or semi-autonomous driving of your vehicle even in situations when GPS is unreliable or not available at all (i.e. indoors, in tunnels, under dense vegetation, etc.).
• Know the position of your vehicle even when no GPS reception is available. That would, for example, enable you to define no-go areas for your vehicle reliably.
• Improve position precision and reliability considerably when compared to GPS-only solutions.
• Improve availability compared to GPS solutions. SLAM is possible 24/7 while GPS reception might be unreliable or not available at all for several minutes or hours every day, depending on GPS satellite positions in your area.
• Create a map of a previous unknown terrain. Use this map for mission planning, terrain analysis or to solve various other problems.

Fig. 1: Computer vision algorithms detect markers that were placed for orientation and one of our SLAM methods measures the marker locations to estimate the current robot position.

Our SLAM technique is based on modern statistical sensor fusion algorithms. To increase the robustness of the SLAM method, we built a flexible software architecture, which allows easy simultaneous use of different sensors. We have years of experience applying CCD cameras, thermal imaging cameras, laser scanners, gyroscopes, accelerometers, GPS, wheel encoders and more. Our powerful computer vision algorithms allow integration of camera images into the mapping process. Modern image analysis methods find interest points for orientation in many situations. Thermal imaging cameras and laser scanners allow mapping even under difficult lighting conditions. For increased robustness, we offer to add markers to your environment, so that an autonomously driving vehicle can find its way even more reliably.

Upon request, our experienced mechatronic engineers will integrate the necessary sensors in your land, sea and air based vehicles. They will develop the appropriate interface, augment the sensors and adapt them to your application.