All the information used by driverless cars comes from sensors. Without sensors, there would be no driverless cars. Sensors are the core part of driverless cars.  Driverless cars can achieve one-second braking, mainly due to its strong perception.  The main sensors of unmanned vehicles include laser radar, millimeter wave radar, computer vision technology, etc., which can test various obstacles and environments, as well as various road signs.  

laser radar

   Laser radar is one of the most powerful sensors in driverless cars.  Its role is very wide, such as: can distinguish real moving pedestrians and character posters, in the three-dimensional space modeling, static object detection, accurate ranging, etc..  The laser itself has very accurate ranging ability, its ranging accuracy can reach the centimeter level.  Laser radar, which transmits laser beams to detect features such as target position and speed, plays an important role in autonomous navigation vehicles (AGVs) and unmanned driving systems because it can form 3D environment maps with accuracy up to centimeter-level.  In most driverless cars, there is a pillar on the roof that attracts attention: the Laser radar.  It can rotate at a high speed of 360°, that is, it can detect pedestrians and obstacles around the car in real time by firing lasers at 360°, and then stop for one second.  

   Laser radar uses technology to calculate the relative distance of a target from itself based on the time it takes for the laser to return to an obstacle.  Laser radar has many advantages: high measurement accuracy, far visible range, good directivity, strong anti-interference ability, and so on. It has the function of many sensors, so it is an important sensor for driverless cars.  However, rain and snow weather is the fatal weakness of Laser radar, because of the working principle of Laser radar, rain and snow weather, frozen ground will cause changes in reflection characteristics, thus affecting Laser radar to build three-dimensional maps.  

Millimeter-wave radar

   Radar working in millimeter wave band detection, usually millimeter wave is in 30 ~ 300GHz frequency domain (wavelength of 1 ~ 10mm) electromagnetic wave, frequency higher than radio lower than visible light and infrared.  The characteristics in this frequency band make it very suitable for application in the vehicle field.  At present, most high-end cars on the market have been equipped with millimeter-wave radar. The most common millimeter-wave radar includes 24GHz band and 77GHz band.  24G vehicle radar is mainly used in vehicle blind spot monitoring and lane change assistance;  Compared with the former, the 77G band radar has a smaller volume and higher frequency, and its radar performance is better than the 24GHz millimeter wave radar. It is mainly used in emergency braking, active car following and other functions with high safety requirements, and has a broad application prospect.  Its huge advantage is high precision, high resolution.  Radar resolution refers to the closest distance between two objects that can be distinguished by radar. High resolution is very critical for the realization and safety of unmanned driving. The resolution of 24Ghz radar is 0.6 meters, and that of 77GHz is 20 centimeters.  

   Millimeter wave radar is very sensitive to velocity and accurate in velocity measurement. It has obvious Doppler effect, and the target velocity can be extracted by detecting doppler frequency shift.  The detection Angle for pedestrians of 10 meters can reach 120°. Although the detection Angle has certain limitations, its advantages, such as higher than visual speed measurement and ranging accuracy, all-weather working ability, small blind area, suitable for rain, snow, haze and sand and other bad weather, are enough to make it win the favor of unmanned driving developers.  

Computer vision

   Camera is the core part of computer vision and the main visual sensor of automatic system.  For driverless cars, it can help determine the position of lane lines and identify vehicles and pedestrians to ensure safety.  Compared with the above two sensors, it has a low price, a large amount of information and good feature recognition. However, it is greatly affected by weather and dark light at night.  And the image recorded through the lens usually has a certain degree of distortion, that is, the existence of image distortion phenomenon, so the application of computer vision is the premise of camera calibration.  

   Different sensors have their own advantages and disadvantages, so it is necessary for a variety of sensors to work together and complement each other to form an advanced driver assistance system, and then realize unmanned driving.