HOME > NEWS > LiDAR Technology based Vehicle Recognition System

LiDAR Technology based Vehicle Recognition System


一、Analysis on vehicle recognition technology

At present, the technology used in vehicle recognition is very extensive. The main modules of vehicle classification are camera, light curtain, 2D scanning LiDAR, 3D solid-state LiDAR and single-beam distance LiDAR. The functions of each technical scheme are limited, and the specific functions are shown in the table below.

Table 1 Vehicle Recognition Demand Table

When using any type of sensor technology, we have to overcome disadvantages of itself and consider construction, police occupied time, maintenance, service life and etc.

二、Overview of LiDAR Vehicle Recognition System

Benewake LiDAR Vehicle Recognition System can automatically match the model information and assist the toll charging by collecting the four key features of vehicle length, width, height and axle. Through the advanced neural network training, our system will output the vehicle type ASAP. The system includes three modules: LiDAR ranging module, control module and installation module. Now it support windows, Linux and other platforms, with rich interfaces and flexible installation.

Figure 1 LiDAR Vehicle Recognition System

Single bean distance LiDAR emits invisible laser beam quickly. When the vehicle is passing by, the beam will returns to LiDAR, and the distance can be measured. Different from the camera, these distance values do not need to be processed as image recognition. As we all know, the camera needs to operate under light environment at night. The actual pictures taken at night are not as "clean" as the academic research pictures. A large number of noise will make the process of image recognition very difficult. In contrast, according to the LiDAR collect data as distance values, the recognition accuracy is higher, and the detection method will be much easier.

Figure 2 Installation sketch map

The LiDAR vehicle recognition system is divided into two parts, as shown in Figure 2.

Part A: Vehicle identification module. By installing the structural parts, two TF03-100 or TF02 Pro are fixed at a 40 ° included angle, one is measured vertically and the other is inclined downward. Usually the module is installed directly on the top of the toll station, under the overpass or under the traffic light pole through the installation hole, and is connected to the main system through the data line.

Part B: Axle identification module (auxiliary vehicle identification, more accurate results) It is composed of a TF03-100 and an installation structure. It is installed on the roadside of the toll station and used for axle statistics. Complement the components of this solution.

Figure 3 TF03 LiDAR Product

As a kind of ranging sensor, LiDAR has absolute advantages in detecting the five data of vehicle length, width, height, speed and axle (used for vehicle splitting), which are mainly shown in the following six aspects:

三、How to use LiDAR vehicle recognition system

After the system is installed on the gantry, the vehicle height information can be directly measured by the LiDAR installed in the vertical direction, and then the vehicle speed can be calculated by triggering two LiDARs. Finally, the contour information of the vehicle can be restored by combining the data measured by vertical LiDAR. The detection flow chart is as follows:

Figure 4 Detection principle flow chart

After the recognition test is completed, the main control board directly transmits the test results to the customer's main control system through the data line. Transmission content includes statistics of vehicle model, vehicle height, vehicle length, vehicle speed and number of vehicle models, see data communication protocol for details.

Figure 5 The original data map of the vehicle models collected by Benwake LiDAR

The detection result is directly output through the data line of the main control board. The standard communication protocol pin contains 18 bytes. By default, it will output the statistical information of the vehicle model, vehicle length, vehicle height, and vehicle speed and vehicle flow. See the table below for details:

Table 2 Description of data communication protocol of Benewake vehicle classification solution

The accuracy rate of the scheme of the vehicle type identification system independently developed by Benewake is 95%. It does not need secondary development and is easy to install.

If you are interested to test, please click Contact Us or send email to bw@benewake.com