High density 3D sensing using a nodding 2D LIDAR and reconfigurable mirrors

Perception in 3D has become a standard practice for a large number of robotic applications. High-quality 3D perception is not easily affordable. Our previous work on a nodding 2D LIDAR provides high-quality 3D depth information. However, its sparse measurements poses challenges in understanding the physical characteristics and kinematic parameters of moving objects within an uncertain environment. The contributions of this paper are two-folds: (1) a novel design of a nodding LIDAR that provides denser scans and achieves a three times higher scan update rate, and (2) a novel dynamic reconfigurability feature of the sensor by allowing a suitable change in its field of view on the fly using a set of mirrors. Additionally, we develop a novel calibration mechanism for this sensor and experimentally validate its effectiveness for dynamic object detection and tracking. The proposed dynamic reconfigurability with denser scans in the field of interest supports the broader applicability of 3D nodding LIDAR in moving target applications or newer applications that demand defining the interest area for perception.