Targeted trajectories optimization from 2D LIDAR measurements

Abstract

In many applications involving robotics and intelligent vehicles, high precision multi-target trajectories are needed for analysis, especially those associated with control and safety applications. This paper presents an optimization approach to obtain multiple targeted trajectories from a stationary 2D LIDAR along the LIDAR's measurement plane. The use of the 2D LIDAR allows this method to work in various environments (e.g. both indoor and outdoor) compared to Global Navigation Satellite Systems (GNSS) and vision based approaches. A cylinder with high reflectivity is attached to each of the targets of interest where the 2D LIDAR measurements correspond to these cylinders are extracted and the trajectories are estimated. This approach is robust toward short duration occlusions from dynamic objects (both targeted and non-targeted) moving in the environment. Preliminary data are presented in the experimental results which includes the processing of multi-target data.