Background:
Light detection and ranging (LIDAR) scanners are non-contact measurement devices that work by emitting a very narrow light pulse and analyzing the reflection of the light pulse from an object. In order to scan a three-dimensional (3D) area from a single observation position, these scanners typically employ actuators to rotate the LIDAR assembly and/or use rotating mirrors to reflect the light pulse around the scene. The result is a 3D point cloud of precise distance measurements of the scene from the observation position of the LIDAR scanner. The device can also map color information from the scene by integrating imaging sensors with the LIDAR ranging sensors.
Summary:
The present invention provides a method for adaptive speed control of 3D LIDAR scanning actuators which provides improved point distribution uniformity and precision of range measurements for rotary 3D LIDAR scanners. The method takes into account the scene being scanned and adaptively adjusts the speed of the rotary actuators by slowing their movement for objects further from the scanner to increase point densities and speeding their movement for objects closer to the scanner to decrease unnecessarily high point densities. Also, the beam divergence is adaptively controlled based on the distance between the scanner and the object and the sensed geometry of the scanned surface. When the geometry sensed consists of noise due to the spot size and/or multiple returns on the photodiode, the focal length is varied to reduce the divergence and increase the precision of the range measurement. In an embodiment of this aspect of the present invention, a tunable focal length soft liquid lens may be used to control the beam divergence.
This invention includes a method of adaptively positioning a scanner within a three-dimensional scene. A computational method is used to identify an optimal next position for the Scanner based on the current position of the scanner and the scan data produced from that point of the scene. The positioning method takes into account both known features and the occluded areas of the scene. It also includes a method of detecting occluded areas of the scene. The method analyses the distance of each data point from the scanner to identify pairs of adjacent points having distances that differ by a threshold value. This aspect of the invention also includes a geometric method for establishing the threshold value based on the scan pattern analysis. The scan pattern analysis incorporates the scanner specifications and operational modes. Occlusion boundaries are estimated based on the positions of such pairs.
An aspect of the this invention includes a method for registering the scan data generated from various scan positions to a comprehensive mathematical model of the scene. Such registration methods are also be referred to as 3D or 4D stitching and 3D panorama generation. Computational methods used in the registration method operate on a four coordinate system: three positional coordinates and fourth coordinate corresponding to the hue of the scene at the measured point. This invention also focuses on a method for determining best scan positions for generating a series of scans based on analysis of the scene in previous scans. Each next scan position is determined by adaptively analyzing the scene in terms of both the area covered by the scan and identification of occluded areas that need to be filled. The next scan position determination methodology consists of algorithms for registration of individual scans into a global map. detection of occluded areas, establishment of frontiers and scanner blind spots and computation of next best vantage points. Other aspects of the present invention include hardware and algorithms for implementing the aforesaid aspects of the invention.
Benefits:
- Scanning hardware and adaptive algorithms for controlling such hardware
Applications:
- Algorithms for generating 3D global maps from Scanning data.
Full Patent: Range Dependent Resolution Control For Three Dimensional Laser Scanning