We are living in a world with fast growing technology, like IT, IC, AI and so on. No exception, autonomous cars have made a huge impact on traditional motor industry. Today we will have an overview on it.
The milestone of autonomous cars was based on Mercedes-Benz robotic Van in 1980. It used vision guided system like LIDAR, radar, GPS, infrared, deep mind, particle filter and computer to sense environment and recognize object.
All these sensors will work as “cars’ eyes” to steel cars on road.
For LIDAR, it can send a beam of laser to object and collect data
from reflection. Because of higher precision, it can collect detailed 3D information of scanned object when it moves with stable angular speed, then it will form “point cloud”, which will help system to locate cars’ position and analyze surrounding environment. Similarly, radar will have the same work with LIDAR, but its precision will be lower than LIDAR.
For GPS system, it usually helps autonomous cars to locate their position in the map by satellite. And steel car to drive in unfamiliar environment.
While for cameras that installed on cars will help car to recognize object around them, which will lead car to avoid obstacles.
Refer to deep mind and particle filter, they are totally based on above sensors, for example, deep mind will combine camera to detect and recognize object, and particle filter is the algorithm that applied in LIDAR.
Let’s have a deep look on LIDAR. LIDAR has three main parts, including launcher, scanner and sensitometer. After scanning the environment, LIDAR will form “point cloud” to memorize objects information. At same time, LIDAR will also record time and Azimuth. Per these message, we can calculate all points ’coordinate.
From the above picture, we can get the x and y coordinates of this point respectively: cos(b)*xy_dist和sin(b)*xy_dist。(xy_dist is the projection of distance on xy surface)
Through the whole process, we can get the exact point position and then using point cloud to form 3D information. Now we know LIDAR will help cars to locate and avoid obstacles, but in real life we have to notice that LIDAR is really expensive, especially for commercial use. And it must adaptive for more complicated circumstance. So we can conclude that LIDAR will be widely using on autonomous car but challenges still along with it.