There are many questions surrounding lidar and how it will influence the future of autos. Auto Futures spoke with lidar experts and analysts all over the world to shed some light on the lidar ‘boom’, such as who needs it, what it is about, when it will be deployed, where it will be used, why it is used and how it will work. Analysts from IHS Markit, Frost and Sullivan, Tractica and BIS Research offer their insights into the fascinating lidar marketplace set for huge growth.
Who invented it?
In 2005, Velodyne’s founder Dave Hall invented 3D-lidar to give autonomous vehicles real-time, 360-degree vision, after competing in the DARPA Grand Challenge for autonomous vehicles. Today, Hall continues to work in the lab with his engineers.
“Velodyne is still the leader for lidar and benchmark testing of vehicles for autonomous driving,” says Egil Juliussen, PhD and director of automotive technology research at IHS Markit.
Lidar is now seeing a rise in demand from developers of autonomous car systems, providers of 3D mapping applications and companies which want to implement ADAS systems into current vehicles.
What is lidar?
“Light detection and ranging sensors (lidar) are used to scan the environment with a non-visible and non-harmful laser beam, which is used to visualize objects and measure range. It also aids in creating a 3D image of a vehicle’s environment,” says Faisal Ahmad, CEO, BIS Research.
In general, the whole lidar system is very complicated; there are many different chips which can send out signals, record and control an autonomous car. There may also be sensor fusion, which intelligently combines data from several sensors to improving system performance. In addition to this, lidars present different benefits such as laser position, frames-per-second and range which, Juliussen says, is why there are so many different companies involved.
However, he adds, lidar is not without its problems. Lidar can be blocked by dirt on the sensors, which is why companies tend to position them inside the window cap or use self-cleaning sensors.
What are the different kinds of lidar?
There are three main types of lidar: mechanical, MEMs and solid-state. These each present different capabilities for autonomous vehicles, such as improved detection, price, size and durability. To help you understand each in more detail, we’ve included a quick summary of the different types of lidar below.
What is mechanical lidar?
Mechanical scanning lidars use rotating turrets and mirrors to direct the laser beam across a field of view. They are usually mounted on the car for the best accuracy and range. You may have seen these devices in the media, which look like a spinning can or bucket on top of autonomous vehicles. According to Ahmad, mechanical lidars are usually larger and more expensive than other alternatives.
Arunprasad Nandakumar, team leader, chassis, safety & autonomous driving systems at Frost & Sullivan, says that many OEMS are still using these mechanical lidars.
What is MEMs lidar?
MEMs-based lidars use microelectromechanical mirrors, which makes them smaller and cheaper than other systems. The laser is fired into a tiny rotating mirror that tilts at different changing axes, instead of rotating like the larger and less agile mechanical lidar.
“By around 2021-2022, there will be great potential for MEMs lidars, driven by organic growth for level 3 and level 4 autonomous driving,” says Nandakumar.
What is solid-state lidar?
Solid-state lidars have no moving parts, making them less susceptible to vibrations. The system is built entirely on a silicon chip. However, due to its non-rotating nature, solid-state lidars face field-of-view limitations. Ahmad still expects these to dominate the automotive lidar market by 2028.
“The gold standard for lidars will be solid-state as long as they can meet ASIL standards (Automotive Safety Integrity Level) and can achieve up to 200 metres range,” says Nandakumar. He doesn’t expect solid-state lidars until 2025.
Juliussen thinks that, when solid-state lidars eventually come out, they may be deployed in advanced driver assistance systems (ADAS). Currently, there is massive fragmentation in the lidar market. Nandakumar says that there is still no benchmark as many companies are still only in the second prototype stage.
When will it be widely available?
Nandakumar predicts that lidar will take off in 2021 when automotive supply companies such as Bosch and Continental start to produce their own systems
But the price of lidar is still too expensive for most companies unless they are generating a lot of revenue from robotaxi businesses. More automakers will start to use lidar sensors once the price drops a few hundred dollars, says Juliussen.
“For lidar to catch on, the price has to fall and self-driving technology has to be popular,” says Anand Joshi, principal analyst/strategy consultant – Artificial Intelligence, Tractica Research.
For the price to come down, the technology has to mature. This isn’t exclusive to lidar systems; all technology surrounding automation has to improve. But the most important factor, says Joshi, is safety.
“The most important thing is that nobody should get hurt or die in the process,” says Joshi, “Other things affecting the future of autonomous driving are regulations and insurance frameworks. In order for lidar to take off autonomous vehicles have to take off.”
“Automotive lidar enables the vehicle to sense its surroundings with highly accurate data when compared to other object sensing components such as cameras, radars, ultrasonic sensors and infrared sensors,” says Ahmad.
Because lidar is important to autonomous driving solutions, it offers automakers a new value proposition.
“Autonomous driving is is a new tipping-point – an added value for automakers because of its uniqueness and the driving forces of three prongs of benefits which are safety, convenience and the hope that it will reduce congestion,” says Nandakumar.
How will it be implemented?
For lidar to work flawlessly, it has to be integrated with artificial intelligence, software and hardware.
“You cannot isolate sensors separately from the integral equation and the software stack for level 3 and level 4 autonomous driving. There is no single solution. It is the combination of good sensors and a good software stack, which is why Tesla develops its system in-house for total control and optimization. Waymo is also doing this,” says Nandakumar.
Nandakumar predicts that lidar usage for Level 4 autonomous driving will first appear in premium vehicles in the $80,000 – $200,000 price range. He says it makes sense for mass-market automakers to develop autonomous robotaxis and sell them by-the-mile as opposed to the entire vehicle to avoid high costs.
However, autonomous companies are still grappling with how to fuse lidar data with data from other sensors and how to make it function perfectly.
“Ultimately, the most important aspect of autonomous vehicle adoption is the human aspect and the answer to a simple question ‘Do I trust my car enough that it will not kill me?’” says Joshi, who personally does not use automated cruise control on his vehicle because he feels that he will not pay attention to the road.