Contineously growing potential of vehicle autonomy reveales great prespectives for increased vehicle tracking and fleet management efficiency, thus increasing safety and cutting off operation costs. Currently, LIDAR (Light Detection and Ranging) based systems seem to form a solid fundament for the image based vehicle tracking. In this review we will briefly discuss LIDAR and it’s potential in fleet management.
What is LIDAR?
LIDAR acronym stands for Light Detection And Ranging. LIDAR was developed in the 1960s, shortly after the discovery of the laser .LIDAR is essentially a laser based system, that aimed at measuring the distance to a target by illuminating the target with laser light and measuring the reflected light by special sensor. Similarly to the radar and sonar, the LIDAR concept is pretty simple, however light and photons different in behavior and management compare to RF or acoustic energy. We will discuss some of the LIDAR working principles further in this review.
LIDAR: working principles
In a rather simplified form, LIDAR workin principle could be described as follows: a system includes a laser diode or LED, a photosensor, a time measurement circuit and actual object as shown on a figure below. There are two ways to measure time of flight (ToF): direct and indirect. In direct mode, short pulses of light are directed to an object, and the time until each pulse returns to the sensor is used to calculate the distance to an object. Indirect ToF utilize a continuous wave of modulated light. A photoreceptor is used to sense the reflections. Afterwards, the timing and phase of the reflected light is used to obtain the distance to the object.
The processing algorithms transform the huge amounts raw reflection data into a volume and vector map relative to the vehicle’s position, speed, and direction. The resultant image insight is used for object identification, motion vectors, collision prediction, and collision avoidance.
After the massive of raw data is generated, advanced processing algorithms are employed to further produce a volume and vector map with respect to vehicle speed, position and direction. The resulting data are employed to identify objects, motion vectors and prevent/predict collisions. Different types of surfaces resulting at various types of reflections, and there are some additional special techniques and algoritms in use to account for that.
In terms of LIDAR measurement procedures, there are two common pathways, that are shown on a figures A and B above. Solid state LIDAR (figure A) is a system built entirely on a chip. No moving parts are involved. In such system, over hundred laser diodes and a large receiver array are required. The laser diodes must be equipped with pulse widths in the nanosecond range and with currents of a few amperes. This technology results at manufacturing challenges, but at the same time has further potential for miniaturisation.
When micro-mirror based on MEMS technology is used, the individual laser beam is radiated and reflected in a line shape. The reflected light is evaluated by a corresponding optics sensor in the SPAD cell. It is a system with movable components. This technology needs significant attention to precision and accuracy, since moving objects may misalign for instance while driving through some rough surfaces.
Some of the most popular modern commercial LIDARs are provided by Velodyne. For example, a HDL-64E LIDAR sensor, designed for obstacle detection and navigation of autonomous ground vehicles and marine vessels. Its featuring durability, 360° field of view and very high data rate, 64 channels and 120m range. This sensor is claimed to be ideal for the most demanding perception applications as well as 3D mobile data collection and mapping applications.
Another example is OS2 lidar sensor series by OUSTER. Depending on the actual model, it may have from 32 to 128 channels with a range of up to 128 meters, and up to 1.5cm precision, with an ability to collect up to 2621440 points per second.
LIDAR systems are seems to be among the key components towards enabling a fully self-drving car. Accompanied with cameras, MDVRs, other types of sensors, and advanced algoritms it provides solid basis for further development in self driving vehicles, advanced ADAS systems, remote sensing and monitoring solutions. LIDAR systems where very bulky and extremely expensive at the very begining and now already developed into relatively compact systems with a price range dropped more than 10 times compare to some initial values. With further cost reduction and minituarisation, the usecases of these systems will contineously expand.
Alireza G. Kashani et al., A Review of LIDAR Radiometric Processing: From Ad Hoc Intensity Correction to Rigorous Radiometric Calibration.2015.
Zhang et al., Image-based vehicle tracking from roadside LIDAR data. 2019.
Cristiano Premebida et al., A Lidar and Vision-based Approach for Pedestrian and Vehicle Detection and Tracking. 2007.