23-11-2020 | | By Robin Mitchell
Recently, Luminar announced that it would be teaming up with Mobileye, an autonomous driving solutions developer owned by Intel. Who is Luminar, what technology does Luminar use, and why is LiDAR critical for autonomous driving systems?
Luminar is a company that specialises in LiDAR technology for use in autonomous driving systems. Founded in 2012 by Auston Russell, Luminar has been able to develop a unique LiDAR solution from the chip level that is, according to Luminar, the only LiDAR solution that meets the requirements for Level 3 through Level 5 autonomous vehicles. These requirements include performance (high-speed moving traffic), safety (ability to detect all objects, vehicles, and pedestrians), and costs.
The LiDAR systems produced by Luminar use the same concept as legacy LiDAR systems, but with a few differences that allow for a low-cost, compact LiDAR system. The laser used in Luminar systems is a single 1550nm fibre laser which allows for extended range (up to 250 meters) and can provide 1,000,000 pulse energy of 905nm while being eye-safe.
Instead of the entire unit rotating, two-axis scanning mirrors which have 120° x 30° field of view rotate and direct the laser beam. This allows for a simpler system with less mass being rotated while also allowing for a more low-profile design. The laser receiver used is one of the most sensitive and highest dynamic range in the world utilising InGaAs technology. However, according to Luminar, this is only achieved when using their in-house ASIC receiver, and their processing system is also in-house designed using a custom mixed-signal ASIC which provides better performance and lower cost than ADC chips.
Recognising the importance of LiDAR in autonomous driving, Mobileye (owned by Intel), has announced their partnership with Luminar to work together on self-driving cars. The agreement between the two companies will see Luminar integrate their solution into Mobileye’s current developments, including their robo-taxi pilot and the first generation of driverless vehicles.
Mobileye, an Israeli subsidiary of Intel, was established in 1999 and started working on visual systems for vehicles that could detect other nearby cars. Their first hardware/software solution, called the EyeQ chip, was quickly used by major car companies including BWM, General Motors, and Volvo (who are particularly known for their outstanding safety record). After years of development and increased sales, Mobileye began to work with carmakers and Intel to produce autonomous vehicles, and in 2017, was acquired by Intel for $15.3 billion.
Mobileye currently plans to have a fleet of self-driving taxis in several regions around the world including Tel Aviv, Dubai, Paris, and Daegu City by 2022. The first systems would cost between $10,000 to $15,000, and by 2025, Mobileye is aiming to bring the cost of a self-driving system down to $5,000. While the COVID-19 pandemic of 2020 has caused widespread economic and social damage, the development of self-driving vehicles could be hastened by the need for social distancing, and the ability to minimise human contact.
There are many technologies that self-driving cars can utilise for sensing their environment, including LiDAR, RADAR, SONAR, and visual. While most, if not all, autonomous systems utilise all of these simultaneously, LiDAR is of particular importance to autonomous driving. RADAR has the ability to penetrate fog, while SONAR can allow for accurate short-range measurements, but LiDAR allows for object detection, distancing recognition, and speed.
While RADAR can also provide information regarding speed and distancing, the wavelengths used are generally too large to produce an image from small objects. Thus it cannot be used for object recognition. The resolution of a LiDAR is much finer, and this allows for object recognition via AI. When teamed up with a visual camera (whereby a received image can be overlaid on a LiDAR map), a system can generate a 3D image of its world allowing it to understand where the road is, where pedestrians are, and what their relative speeds are.
However, a LiDAR based on legacy systems with a rotating dome may be easier to construct, but they are not entirely appropriate for use in autonomous driving systems; this is why the system developed by Luminar is critical. LiDAR systems used in autonomous systems may also greatly benefit from MEMS whereby microscopic mirrors can be manipulated on a chip, thus entirely removing the need for a motor or any moving mechanical parts. Such a system will suffer fewer mechanical issues, have an increased life-span, and may even provide greater mitigation against vibration and impacts as the vehicle travels.