7 Defects in Point Cloud That LiDAR manufacturers won’t tell you


Table of Contents

LiDAR Point Cloud

LiDAR technology has revolutionized the way we perceive and interpret our surroundings. LiDAR has been integrated into various fields, from autonomous vehicles to construction sites, and even archaeological digs. The technology uses lasers to create 3D point clouds, which can then be used for mapping and modeling. However, not all point clouds are created equal, and there are some defects that LiDAR suppliers may not disclose. In this blog post, we will discuss common point cloud defects that are often overlooked by LiDAR suppliers.

According to different ranging methods, LiDAR can be divided into two categories: ToF (Time of Flight) and FMCW (Frequency Modulated Continuous Wave). ToF is currently the most widely used method for mass-produced LiDAR.

“Tail” Problem

In the ToF method, the LiDAR transmitter emits a pulse, which hits an object and returns, and the receiver calculates the time difference between the two when receiving the echo, and measures the distance between objects by multiplying it by the speed of light.

Ideally, the pulse hitting the object’s surface is an ideal spot, but due to the actual pulse having a certain divergence angle, it will hit the object’s surface as a face, and as the distance increases, the face will get larger. This creates a possibility that when there are two objects in front and back, and the LiDAR pulse hits the edge of the front object, it may hit the back object partially. This is known as the “tail” problem of the LiDAR.

Fig1. Pont Cloud-Tail
Fig1. Pont Cloud-Tail


The direct consequence of the “tail” problem is that one pulse sent out by the LiDAR returns two echoes, causing the LiDAR to become “confused” and unable to determine which distance to use. To solve the “tail” problem, the solution would be to use laser pulse transmitters with more focused energy and smaller divergence angles. Alternatively, optimizing the algorithm could be done by judging whether the angle threshold is within a reasonable range, thereby achieving tail point screening and deletion.

Blind Spot “Dead Points”

LiDAR detectors generally have a dead time of several to tens of nanoseconds. Dead time refers to the shortest time required to receive a new laser pulse after receiving one laser pulse. When a laser pulse is emitted, an internal reflection signal is first generated in the laser emission lens and received by the detector. If the obstacle is too close, the pulse-echo of the close-range object cannot be detected because the laser receiver is still in the dead time period, which leads to inaccurate ranging of close-range objects.

The problem of inaccurate ranging of close-range objects detected by LiDAR is called “dead points,” which is a difficult problem that plagues the entire industry and requires continuous evolution of the underlying detector hardware. The close-range area with inaccurate ranging is usually set as a “blind spot,” and the size of this blind spot is usually between 0.1 to 1 meter.

Fig2. Pont Cloud-Blind
Fig2. Pont Cloud-Blind


High-Reflectivity “Ghosting”

For highly reflective objects, after entering the field of view and ranging range of the LiDAR, the output point cloud not only forms an image at the real position but also easily forms a false image in another location with a similar shape and size, which is called “ghosting.” The trajectory of “ghosting” formation differs depending on the type of LiDAR.

Fig2. Pont Cloud-Blind
Fig3. Pont Cloud-Blind

The formation of “ghosting” is due to the fact that LiDAR is very sensitive to the high-intensity echoes reflected by highly reflective objects. In actual driving scenes, common highly reflective objects include traffic signs, cones, triangular signs, license plates, and taillights. As shown in the figure below, the left side was originally unscannable by the laser, but due to the “ghosting” phenomenon of the high-reflective sign on the right, a point cloud with a similar shape and size is formed on the left.

High-Reflectivity “Expansion”

For highly reflective objects, another abnormal phenomenon is the “expansion” effect. …

To address the problem of “ghosting” caused by highly reflective objects, one solution is to use multiple laser beams with different wavelengths, which can help to distinguish between real reflections and ghost images. Additionally, some LiDAR manufacturers have developed software algorithms that can filter out ghost images.

Interference from other LiDAR systems or light sources

LiDAR sensors can be affected by interference from other LiDAR systems or even from other sources of light, such as the sun or headlights from other vehicles. This interference can cause inaccuracies in the LiDAR data and even lead to a complete loss of data.

To address this problem, LiDAR manufacturers have developed different methods for filtering out interference. For example, some LiDAR sensors use polarization filters to reduce interference from sunlight, while others use frequency hopping to avoid interference from other LiDAR systems.


In conclusion, while LiDAR technology has advanced significantly in recent years, there are still several challenges that need to be addressed to improve the accuracy and reliability of LiDAR systems. LiDAR manufacturers continue to work on developing new hardware and software solutions to overcome these challenges and make LiDAR more effective technology for various applications.


Related articles:

  1. How to Get a Lidar Sensor at the Lowest Price
  2. LiDAR industry chain interview : About FMCW OPA, Flash LiDAR
  3. The cost and advantage: About MEMS & Rotating mirror LiDAR


Request User Manual

Request Datasheet