With the breakthrough of computing power and more accessibility of high-performance computers, as well as the technical advances of various sensors, artificial intelligence is being revived.
An epidemic has also accelerated the process of various unmanned works.
  • What is L4 autonomous driving?
    The Society of Automotive Engineers (SAE) has defined 6 levels of autonomous driving - L0 (fully manual) to L5 (fully autonomous), and the U.S. Department of Transportation has adopted these standards.
  • L4 highly autonomous driving :
    L4 vehicles, with an "environmental detection" function, can make rational decisions on their own, such as driving past a stationary object. In most cases, L4 vehicles require no manual intervention, but the drivers can still take over driving. L4 vehicles can travel in autonomous driving mode, which can only be allowed in certain areas such as private residential areas, industrial areas, and large warehouses until the laws and infrastructure are perfected.
    PTN is currently continuing
Brief Comparison of Current Automatic Navigation Solutions PTN Solution



TESLA Factory AGV Solution Food Delivery Solution in Hotels and Restaurants Lobby Solution
Application Scenarios Can work in hotels, hospitals, factories, warehouses, closed communities, parks, and other indoor or outdoor environments All scenarios, relying on high-definition maps Currently it travels on highways, without relying on high-definition maps AVG robot can only work in the factory indoor environment Robots can only work in an indoor environment, and the environment is a small structured environment, such as hotel corridors. Robots provide information services in indoor lobbies such as in hospitals and hotels
Camera Single depth camera and single track camera Multiple HD cameras Multiple HD cameras Some AGVs are equipped with HD cameras Most are equipped with HD cameras Most are equipped with HD cameras
LiDAR None Multiple outdoor LiDARs, at present too expensive for mass production None Single-line or multi-line LiDAR for indoor use Single-line or multi-line LiDAR for indoor use Single-line or multi-line LiDAR for indoor use
RTK is required General GPS None None None
Millimeter-wave radar None
Multiple millimeter-wave radars Multiple millimeter-wave radars None None None
Sonar Optional None Multiple sonars Some are equipped with sonars Some are equipped with sonars, or replace LiDAR with multiple sonars Some are equipped with sonars, or replace LiDAR with multiple sonars
Floor tape None
Most need to be pre-laid with floor tape Some need to be pre-laid with floor tape None
Running speed Low speed, the current maximum speed is 14km per hour Both high speed and low speed are available Both high speed and low speed are available Low speed Low speed
Low speed
HD map (centimeter level) Not required, self-built 3D map (route mapping) Required Not required Not required Not required, general 2D map Not required
Algorithm complexity High High
Mass production
No Yes
Comments A 10-year warranty is provided for 3D cameras, and the whole solution is reproducible, and suitable for mass production Active transmitting equipment, such as LiDAR and millimeter-wave radar, are consumable parts and have a limited lifespan, for example, only one year warranty is provided for velodyne
In actual applications, mistakes may occur in road conditions with complex lane lines, such as sharp corners.
Introduction of current
commonly-used sensors
  • LiDAR
  • Depth camera
  • High-precision satellite positioning
  • Millimeter-wave radar
  • Sonar
Simple comparison of commonly-used sensors
Sensor Price Degree of Difficulty for Algorithm Current Development Advantages Disadvantages Summary

More than RMB 10,000 will be spent for a single-line LiDAR with better accuracy.

Multi-line LiDAR is very expensive, ranging from RMB 10,000 to RMB 100,000

Easier More mature

  • High accuracy, and accurate terrain judgment
  • Fast R&D progress: There have been already many successful cases

  • Expensive
  • Millimeter-level LiDAR is too sensitive to environmental changes, with a high maintenance cost
  • LiDAR itself has losses and needs to be updated frequently
  • Considerably affected by weather and environment (such as natural light, smog).
  •  LiDAR as the main solution also requires high-definition maps

Unable to obtain image data, such as ground road signs, etc.

The development of LiDAR for indoor use is relatively perfect, and it is applicable to indoor, small, regular, and light-stable indoor enclosed environments.

LiDAR for outdoor use is expensive.

Can expect domestic LiDAR to reduce development costs

Depth camera


Below RMB 10,000

High difficulty Larger development space

  • Low cost
  • Can process color data
  • Low dependence on camera types
  • Can be used both indoors and outdoors

  • Heavily rely on computer vision technology
  • High requirements for computer computing power

Can be used in a variety of slow speed indoor and outdoor environments

Low cost but heavily relying on computer vision technology

The technical breakthrough is mainly in computer vision technology, which can be self-developed

High-precision satellite positioning

  • Expensive
  • High-precision satellite Antenna

  • High-precision satellite


Easier More mature

  • Relatively mature technology
  • Stable operation on designated routes that meet the requirements

  • Relying on RTK base station. Unable to receive high-precision satellite signals outside the working range of base station
  • Satellite/RTK signals are susceptible to interference from obstacles such as buildings and trees.

The technology is mature but relies on high-precision satellite base stations, and the base station cost is very high
Millimeter-wave radar Medium Easy Mature

  • Very mature related technology
  • Quick response

Low resolution of radar sonar Can be used as an assistive sensor, not suitable as a core technology
Sonar Cheap Easy Mature

  • Very mature related technology
  • Quick response

Short detection range of sonar Can be used as an assistive technology, not suitable as a core technology
  • Sensor Introduction
  • Depth camera
  • LiDAR
Sensor Introduction
Sensor Introduction
At present, most of the finished automatic navigation systems take LiDAR as the core of operation, and computer vision is only an assistant. LiDAR is indeed an appropriate choice for indoor use only, but the core technology is actually subject to LiDAR manufacturers who are easy to be substituted if they have no independently-developed LiDAR.
Depth camera
Depth camera
Simulate the human field of view, and obtain the actual distance and 3D structure of the object through calculation. Both eyes form a 120°field of view by means of the dual-camera coordination algorithm.
  • 01
    Single-line LiDAR
    Single-line LiDAR
    also known as planar LiDAR, can obtain a cross-section of the actual 3D environment through a 360°scan.
  • 02
    Multi-line generates a 3D environment
    Multi-line generates a 3D environment
    The 3D structure of an obstacle is formed by multiple lines
Get more solutions