Understanding LiDAR technology for ADAS and Autonomous Vehicles

Date: May 7, 2019

Author: Akshay Srinivasa

Autonomous machines are all set to ply on the streets in near future. LiDAR is one system that is aiding autonomous machines to map their surrounding and detect objects on their path. In this article we explore in brief a LiDAR system for object detection application in autonomous machines. Some researchers forecast more than 69 million LiDAR units will be shipped by 2026.
What is LiDAR?
LiDAR stands for light detection and ranging, LiDAR uses pulsed laser (form of light) to measure various parameters in their path. LiDAR facilitates in mapping the surroundings. A typical LiDAR based system of autonomous machine consists of following components-
Component of a Lidar system

Figure:1- Components of LiDAR system

  • Laser source- Eye-safe laser source of wavelength 1550nm are commonly used in automotive application, but these might also cause some damage to certain parts of retina. Ideally less than 1000nm (600nm-1000nm) wavelength laser source is used.
  • Scanning mechanism-The ability of LiDAR system to map the surroundings accurately completes depends on the scanning mechanism. This provides crucial information on azimuth and elevation of the object.
  • GPS and inertia measurement unit- This provides details pertaining to the position of the autonomous vehicle along with orientation of the sensor.
  • Computer- With vast amount of data being received by the scanning system, every pulse needs to be processed to get the precise azimuth and elevation levels of the objects being mapped.
Types of LiDAR based on beam steering
There are various types of LiDAR system in general that are used in mapping, however in automotive domain or autonomous cars it is generally sorted into to category depending on photon emission; ¬Electro Mechanical- Uses optics that are being rotated continuously to provide a 3600 view. These systems are generally bulky in nature and sometimes costs more than the actual vehicle implementing this. ¬Solid state LiDAR- Electro mechanical LiDAR relies on moving parts (Rotating continuously) With involvement of moving parts efficiency of the overall system is bound to reduce over time. Solid state LiDARs are built on silicon chip and have no moving parts, mechanical rotation is not required.
Working of LiDAR system
People familiar with autonomous cars would have seen a dome like device on top of a car revolving. This dome like device is a LiDAR sensor that uses pulse laser to measure various parameters around a car just like a radar-based system or a sonar. A LiDAR system uses pulsed laser which is emitted on to a surface, the reflected pulse is then processed to identify various objects in the path, it is safe to say this is the reason for the name light detection and ranging.
Working of LiDAR based object detection

Figure:2 Working of LiDAR based object detection system

A typical LiDAR system fires around million pulses each second. One of the key parameters to be considered before using the data of reflected pulse is the height of the unit and the velocity of the vehicle which houses the sensor so as to get an accurate estimation of mapping. LiDAR detection in general practise is classified in to two categories namely; coherent and incoherent.
  • Coherent detection- The scanning mechanism that receives the pulses after reflection uses ToF (Time of flight) to determine azimuth and elevation.
  • Incoherent detection- The scanning mechanism compares the received pulse with that of the transmitted one to determine azimuth and elevation.
Each of the reflected pulses irrespective of detection method are processed for a 3D object detection using point cloud processing techniques.
Advantages of using a LiDAR system for object detection
LiDAR has numerous advantages thanks to its working way of using laser pulses.
  • ­Object detection from few mm to all the way up to 200m
  • ­Low spatial resolution of 0.10, this helps in differentiating objects that are close to each other but far off from the LiDAR unit.
  • ­LiDAR has better azimuth and elevation field of view thanks to 3600 rotation compared to a radar-based system
  • ­LiDAR systems have high operating efficiency in low light condition (darkness) than any camera-based object detection system
LiDAR in automotive industry
Autonomous systems are here and we will see them plying on roads in the next decade, from delivery robots to shuttle services and on demand car services. LiDAR in automotive industry is used for surrounding mapping. With information of the surrounding, a CNN model can infer on any objects surrounding the autonomous machine and plan its path without colliding with other objects or vehicles on the road. Apart from obstacle detection LiDAR is also used in localization-SLAM (simultaneous localization and mapping). This is application of LiDAR in automotive/ autonomous machines is a method of identifying precise location of an object and simultaneously tacking it.

Figure:3 LiDAR in autonomous car

Conclusion
Even though there is an argument that LiDAR based systems are still a long way to be commercialised for autonomous system, advantages of using a LiDAR system makes it even more worth while to develop it. LiDAR can be used with other sensors like camera and radar which assist the LiDAR system to increase the accuracy even further.
Further reading
Interested in LiDAR ADAS application? Click here
Interested in Radar Automotive application?
Here is our radar series of blogs explaining Automotive Radar Interested in Radar service and Radar SDK (Jump start solution)?

You might also like these blogs

post
IoT in Agriculture – A Smart Solution for All Concerns in Food Production

Intelligent, connected devices are on the rise in every sector,…

Read More
post
The Evolution of Smart Home

Any device in your home that uses some amount of…

Read More
post
A brief analysis of Indoor Positioning market

As we progress, we are no longer restricted to getting…

Read More