A visual study of ICP variants for Lidar Odometry

Odometry with lidar sensors is a state-of-the-art method to estimate the ego pose of a moving vehicle. Real-world effects such as dynamic objects, non-overlapping areas, and sensor noise diminish the accuracy of ICP. We build on a recently proposed method that makes these effects visible by visualizing the multidimensional objective function of ICP in two dimensions. We use this method to study different ICP variants in the context of lidar odometry. In addition, we propose a novel method to filter out dynamic objects and to address the ego blind spot problem. Keywords Iterative closest point Registration Odometry Mapping Lidar odometry 1 Introduction Odometry is a widely used technique to estimate the trajectory of a navigating robot or an automated driving system (ADS). As more and more ADSs make their way to higher automation levels, the demand for higher accuracy and robustness rises accordingly. One example of this demand is the use of high-definition maps in such systems. To leverage the map information, the AD system first needs to localize itself in the map. Therefore, the accuracy of the localization influences the overall confidence that can be placed in the map information. Since odometry plays an essential role in many localization systems, the robotics community is interested in improving its accuracy and robustness.