What do the laser radar and camera in the driverless technology do?

Lidar and camera performance comparison

In unmanned environment-aware devices, lidars and cameras have their own advantages and disadvantages.

The advantage of the camera is that the cost is low, and there are many people who use the camera to develop algorithms, and the technology is relatively mature. The disadvantage of the camera, first, it is very difficult to obtain accurate three-dimensional information (single-eye camera is almost impossible, some people have proposed binocular or tri-camera to do); another disadvantage is that it is relatively limited by ambient light.

The advantage of the lidar is that it has a long detection range and can accurately acquire the three-dimensional information of the object; in addition, its stability is quite high and the robustness is good. However, the current cost of laser radar is high, and the final shape of the product has not yet been determined.

For both sensor applications, both the camera and the lidar camera can be used for lane line detection. In addition, laser radar can also be used for tooth detection. For license plate recognition and the identification of both sides of the road, such as speed limit cards and traffic lights, it is mainly done with a camera. If the obstacle is identified, the camera can easily classify the obstacles through deep learning. However, for Lidar, it can only be divided into a large class of obstacles, but the judgment of the motion state of the object is mainly done by laser radar.

Multi-line lidar----How many lines are suitable?

At present, there are not many manufacturers of laser radars in foreign countries and in China. For example, Velodyne introduces 16-wire, 32-wire and 64-line laser radar products. Quanergy's early 8-line laser radar product M-8 (Solid Lidar is under development). Ibeo mainly introduces 4-wire laser radar products, mainly for assisted driving. RoboSense is introducing a 16-line laser radar product.

How many lines of laser radar products can meet the needs of unmanned manufacturers, including traditional car manufacturers and Internet car companies?

Multi-line laser radar, as the name implies, is a scan of multiple strands by the rotation of the motor through the distribution of multiple laser emitters in the vertical direction. How many lines of lidar are appropriate, mainly to say how many lines of laser radar swept out of the object can fit the needs of the algorithm. In theory, of course, the more and more dense the wire harness, the more complete the description of the environment, which can also reduce the requirements of the algorithm.

The industry generally believes that 64-line laser radar products like Google or Baidu are not the final product form of Lidar. The direction of Lidar's products must be miniaturized, and the vertical resolution of the transmitters between two adjacent sources must be continuously reduced to achieve higher wiring harnesses.

Lidar product parameters include four aspects: measurement distance, measurement accuracy, angular resolution, and speed of laser single-point emission. I mainly talk about resolution: one is vertical resolution and the other is horizontal resolution.

The horizontal viewing angle of the multi-line lidar is now 360 degrees, and the vertical viewing angle is the vertical viewing range. The resolution is very similar to the pixel of the camera. The laser beam finally forms a 3D laser point cloud, similar to an image with many pixels. The denser the laser point cloud, the more comprehensive the perceived information.

It is not difficult to achieve high resolution in the horizontal direction, because the horizontal direction is driven by the motor, so the horizontal resolution can be made very high. At present, the laser radar manufacturer's products have a horizontal resolution of 0.1 degree.

The vertical resolution is related to the geometry of the emitter and is also related to its arrangement. The smaller the interval between two adjacent emitters, the smaller the vertical resolution will be. It can be seen that the increase of the wire harness is mainly to describe the same object more fully. If you do not increase the wiring harness by reducing the vertical resolution, it does not mean much.

How to improve vertical resolution? At present, the industry is realized by changing the arrangement of laser emitters and receivers: the denser the row, the smaller the vertical resolution can be. On the other hand, through the coupling of multiple 16-line lidars, the overall reduction of vertical resolution is achieved without increasing the vertical resolution of a single lidar.

However, both methods have certain drawbacks.

In the first method, if the wire harness is added without increasing the vertical viewing range, there is a certain ceiling. Because the geometry of the laser emitter is difficult to shrink further, for example, to achieve a vertical resolution of 1 degree, if you want to achieve 0.1 degrees, it is almost impossible.

The second method, multi-sensor coupling, that is, multiple lidar coupling, because it is not a single product, will have high requirements for future calibration.


Safety Light Curtain

SDKELI has been manufacturing safety light curtains for more than 20 years. Its brand [SHANGSHOU" and [SDKELI" owns largest market share in China because of reliable quality and proper price. All the safety light curtains are type 4 with CE certificate.

Safety Light Curtain,Safety Curtain,Laser Safety Light Curtain,Safety Optic Light Curtain,Security Light Curtain,Press Brake Safety Light Curtains

Jining KeLi Photoelectronic Industrial Co.,Ltd , https://www.sdkelien.com