Often, the SLAM problem is viewed as one of the keys to truly realizing self-driving robots. Therefore, the SLAM problem is also the core technology of self-driving cars. However, many problems make it impossible for SLAM algorithms to drive vehicles for hundreds of kilometers under very different conditions. There are two main problems dealing with SLAM for self-driving cars: (1) localization drifts over time and (2) maps may not necessarily satisfy every driving condition. The former problem is well known in the SLAM community that as the distance travelled by the car increases, the localization estimate given by the SLAM algorithm will deviate the car from the true trajectory; without prior knowledge, correct localization is not even guaranteed within a few kilometers. The second question, regardless of the conditions, needs to have a map sufficient for the positioning task. In this chapter, we mainly see how SLAM is being applied to autonomous driving and then we focus on the high-definition map creation.
SLAM in Autonomous Driving
Autonomous driving algorithms and Its IC Design ; Kapitel : 5 ; 127-152
10.08.2023
26 pages
Aufsatz/Kapitel (Buch)
Elektronische Ressource
Englisch
DL-SLAM: DIRECT 2.5D LIDAR SLAM FOR AUTONOMOUS DRIVING
British Library Conference Proceedings | 2019
|DL-SLAM: Direct 2.5D LiDAR SLAM for Autonomous Driving
IEEE | 2019
|Autonomous driving trolley device based on multi-sensor fusion slam
Europäisches Patentamt | 2024
|ACCURATE ELEVATION MAPS BASED GRAPH-SLAM FRAMEWORK FOR AUTONOMOUS DRIVING*
British Library Conference Proceedings | 2019
|