How does a Lidar sensor work?

The more comprehensive and reliable an autonomous vehicle’s perception of its surroundings needs to be, the more crucial the number and arrangement of sensors, as well as the corresponding data processing by software algorithms. Lidar sensors play a central role in environmental detection. But what exactly do these systems do? How does Lidar work? When will it be implemented in production vehicles, and how will the technology evolve in the future? automotiveIT provides answers to key questions.

What Is Lidar?

How good is the VW ID.7 in winter?

Volkswagen's flagship is being showered with praise. But how good are the assistance systems of the electric successor to the Passat, and what does the car offer in terms of user experience and infotainment? We conducted a practical test.

Timo Gilgen

Light Detection and Ranging, or Lidar, is a system that generates high-resolution 3D information in real time. Unlike radar, Lidar sensors perceive their environment solely through light, using a photo sensor. The system calculates the time it takes for an emitted laser beam to hit an object and reflect back. By sending out a multitude of signals, a 3D map of the surrounding area is created.

Beyond its use in autonomous vehicle assistance systems, Lidar is also utilized for terrain mapping and building measurement. In consumer electronics, Apple has been incorporating its own Lidar systems since the iPhone 12 Pro, aiming to enhance the precision of augmented reality tools. The origins of Lidar systems date back to aerospace applications, where they have been used since the 1960s.

What Lidar Technologies Exist?

Current Lidar systems mainly differ in their design (Spinning, Scanning, or Flashing Lidar), their measurement method for environmental detection (ToF or FMCW), and their laser wavelength (850 nm, 905 nm, or 1,550 nm). Another distinction is whether the sensor system has moving components—Lidar systems without moving parts are classified as solid-state Lidar. Additionally, different Lidar systems vary in terms of field of view and range. The average range of current systems is about 250 meters, depending on the reflectivity of detected objects. However, some systems on the market can detect distances of up to 500 meters.

• Design: Spinning vs. Flashing Lidar

The distinction between Flashing Lidar and Scanning or Spinning Lidar relates to their core operating principle in environmental detection. Flashing Lidar covers a specific area of the environment with a single light emission, capturing it in its entirety. Spinning Lidar, on the other hand, features rotating mirror systems that emit laser beams in multiple directions, thus mapping the entire environment by covering multiple sections.

Due to their moving parts, Spinning Lidar systems are often more expensive, require more space, and are more susceptible to environmental influences such as vibrations or temperature changes. An alternative is the use of MEMS mirrors (Micro-Electro-Mechanical Systems), which provide a smaller, more durable solution without moving parts. In Lidar systems, these microelectromechanical mirrors enable the emission and reception of laser beams in different directions. The chip-based solid-state design of the system reduces both costs and space requirements for installation. However, due to relatively small mirror elements, range and signal quality still have room for improvement. For example, Continental plans to begin producing a high-resolution solid-state Lidar with a range of approximately 300 meters in 2024.

• Measurement Method: ToF vs. FMCW

Currently, there are two different methods for Lidar environmental detection: Time of Flight (ToF) and Frequency-Modulated Continuous Wave (FMCW).

The FMCW method focuses on frequency modulation, meaning a continuous laser signal is emitted with constantly changing frequencies. The returning light is compared with the emitted frequency to calculate the distance to detected objects.

The ToF method, on the other hand, uses time-of-flight measurement. The distance to an object is determined based on the time it takes for laser pulses to return to the sensor after being emitted. To measure the speed of approaching objects, the ToF method requires multiple data points, whereas FMCW Lidars can gather this information instantly.

Is the FMCW method superior to Time of Flight (ToF)? "That cannot be said in general terms. Both methods have strengths and weaknesses," explains Mathias Müller, CEO of Lidar provider Blickfeld. "However, what ToF has over FMCW is its technological maturity." ToF has been successfully used in Lidar sensors for years, whereas FMCW is still in its early stages.

This assessment is shared by Clément Nouvel, Lidar Technical Product Line Director at Valeo. In terms of market penetration and cost, FMCW-based solutions are not expected to be available in mass production until 2027.

Even commonly cited advantages of FMCW—such as a better signal-to-noise ratio—remain controversial. Lidar provider AEye challenges many claims about FMCW in a comprehensive whitepaper. It states that claims about FMCW being more efficient, faster, offering greater range, and delivering clearer signals than ToF systems have not been proven or are outright false. The report concludes that high-performance, agile-scanning ToF systems are better suited to meet the requirements of autonomous driving in terms of cost, range, efficiency, and point cloud quality compared to FMCW.

• Wavelength: 850 & 905 nm vs. 1,550 nm

Another debate in Lidar technology revolves around laser wavelength. The industry standard currently stands at 905 nanometers, but 1,550 nanometers is another option.

Advantages of 1,550 nm Lidar is that they are less susceptible to sunlight interference and there is reduced risk of causing eye damage to humans.

However, 1,550 nm sensors are more expensive and complex to develop. Due to their higher energy output, they overheat faster and can also damage the environment, such as smartphone cameras if exposed to their laser.

At present, low-cost diode lasers with high short-term pulse power are more readily available at 905 nm. Due to different semiconductor materials, diode lasers at higher wavelengths have lower pulse power and are more expensive. Other approaches for higher power output, such as fiber lasers, are even more complex and costly.

Nvidia Delivers New Computing Power for Autonomous Driving

When it comes to the advancement of autonomous driving functions, no one has been able to bypass Nvidia's supercomputing platforms for a long time. At this year's GTC, the tech player announced a series of new and deepened partnerships.

Yannick Tiedemann

What Are the Disadvantages of Lidar Sensors?

Nearly all automakers rely on a triad of environmental sensors – cameras, radar, and Lidar – in combination with high-definition mapping data for autonomous vehicles.

The main reason for redundancy in sensor systems is their individual weaknesses:

• Lidar, like optical cameras, is susceptible to obstructions such as fog or heavy rain and requires high data processing power.
• Unlike cameras, Lidar sensors can only detect the shape, distance, and movement of objects but cannot recognize traffic signs.
• Radar systems, while less affected by visibility conditions, depend on the reflectivity of object surfaces for accuracy.

"Automakers simply want a camera, a Lidar, and a radar together," explains Erich Smidt, General Manager of Innovusion Europe, in an interview with automotiveIT. "Each technology operates on a different frequency." This sensor triad is expected to persist for the next 10 to 20 years.

Whether Lidar will still be needed for autonomous driving in the more distant future is therefore a question of redundancy. Autonomous driving at SAE Level 3 or higher will not work without Lidar, explains Joachim Mathes, CTO of the Comfort & Driving Assistance business unit at Valeo. However, Lidar systems could face competition from 4D imaging radar systems, which are significantly more robust and cost-effective to implement.

The only car manufacturer currently taking a different approach to autonomous driving sensor technology is Tesla. For a long time, the U.S. automaker’s Autopilot system relied on neither Lidar nor radar. Recently, however, this has changed, and Tesla has returned to using radar systems. According to insiders, the company is also closely monitoring 4D radar technology.

Can 4D Imaging Radar Replace Lidar Technology?

The question of whether Lidar is truly necessary resurfaces periodically. Tesla is known for its unique stance—trying to rely solely on cameras and avoid Lidar and radar entirely. However, Tesla has not proven the feasibility of this approach at higher levels of automation.

Recently, the discussion has shifted toward 4D Imaging Radar—a system that simultaneously detects distance, speed, direction, and height of objects. Its main advantage is that radar technology is significantly cheaper than Lidar.

"The potential is there," says Jeremy Carlson, Associate Director of Autonomy at S&P Global, "but it has not yet been demonstrated that 4D Imaging Radar can completely replace Lidar."

Pierrick Boulay agrees: "4D Imaging Radar currently has a maximum resolution of 1 degree, while Lidar already achieves 0.05 degrees—with further improvements expected." He does not see 4D Imaging Radar as a Lidar replacement for budget vehicles, as "redundancy beyond cameras and traditional radar would be too expensive."

Who Are the Key Players in the Lidar Market?

According to data from the Yole Group, the global Lidar market had a modest volume of $317 million in 2022, marking a 95% increase compared to the previous year. The largest share of this market belongs to the Chinese supplier Hesai, which increased its share by five percentage points to 47% within a year. The company stated that it had delivered over 100,000 Lidar units by the end of 2022. Yole estimates that Hesai has eleven OEM customers, six of whom planned to begin vehicle deliveries by the end of 2023.

The second-largest player in the market is the U.S.-based Innovusion, with a 15% market share. For example, the Lidar technology in Nio's ET7 sedan comes from Innovusion. The third-largest company, Valeo, has lost significant market share over the past year and now holds 13%. By mid-2022, Valeo had produced nearly 200,000 units that were in use on the roads. The company secured Stellantis as a customer last year. In March 2023, Valeo announced that its latest Lidar product, Scala-3, had received orders worth €1 billion. Starting from 2022, Yole expects the entire automotive Lidar market to grow by 55% annually, reaching $4.5 billion by 2028.

Among the top three European automotive suppliers, only Continental remains active in the Lidar business after ZF and Bosch exited the technology. In 2020, Continental invested in the California-based sensor pioneer AEye. Its long-range Lidar was originally scheduled to enter mass production in 2024. When asked if the company was sticking to this timeline, Continental stated that discussions with various customers were ongoing. Additionally, the company already has a short-range Lidar in mass production for a premium OEM and another short-range Lidar for small city cars, where it is used as a city emergency braking assistant.

New IAMTS Chairman Aims for Global Harmony

Mr. Quernheim discusses his mission to unify global standards in autonomous driving and enhance safety through collaboration.

Andrea Hoffmann-Topp

Where Is the Lidar Market Headed?

Pierrick Boulay, senior technology and market analyst for photonics and sensors at Yole Group, predicts that the Lidar market will change dramatically by 2035.

"It will follow the same pattern as radar and cameras, where today only four or five players control 70-80% of the automotive market," Boulay explains. Smaller Lidar companies will initially struggle with low market shares.

"I expect further acquisitions, possibly even by established automotive suppliers."

So far, mergers and acquisitions have been slow, but some recent examples include Microvision acquiring Ibeo and Velodyne merging with Ouster. Due to the challenging economic conditions, some young technology firms may seek acquisitions or alternative business models to survive.

One alternative strategy for emerging Lidar companies is to shift focus to other industries where quality standards are less stringent, and Lidar is used in controlled environments, such as intralogistics and industrial automation. Some Lidar manufacturers now present the automotive industry as just one of many potential markets. However, Jeremy Carlson believes there is still hope for smaller tech firms: "If there's one segment of the automotive sensor market where small tech companies still have a chance, it's in Lidar—rather than the already mature camera or radar markets."

What Role Does Software Play in Lidar?

The point cloud generated by Lidar provides a snapshot of objects in the environment at the moment of measurement. To interpret this data and make it usable for autonomous vehicles, advanced algorithms are required.

Software has two key functions:

1. Labeling the detected information
2. Sensor fusion, which combines data from multiple sensors to create a unified environmental model

For example, in 2024, Valeo plans to launch an algorithm that classifies road users and other objects while predicting their movement based on unique assigned IDs. Since data processing typically relies on AI systems, training these algorithms requires significant resources and effort.

Which Cars Use Lidar?

The Lidar market is still in the development phase rather than mass rollout. According to Thomas Luce, Germany CEO of Microvision, the technology will likely remain exclusive to luxury vehicles until 2027."By 2028, production volumes could be high enough to significantly lower costs—possibly down to around $600 per unit."

Currently, Lidar sensors cost around $1,000 each. However, the industry expects a massive market expansion in the coming years, potentially driving costs down. Valeo estimates the market will reach $50 billion by 2030.

So far, Lidar technology has yet to be widely integrated into production vehicles. Some early adopters of Lidar or companies planning integration soon include Mercedes-Benz, Honda, Volvo and Polestar.

Where Is Lidar Technology Headed?

The future of Lidar technology is moving toward solid-state solutions, which are less susceptible to environmental factors. Currently, the industry is working to develop single-chip approaches, but practical solutions are still a few steps away.

"We believe that the promises of FMCW will be fulfilled once we can integrate everything onto a single chip with just a small number of additional components," says Valeo expert Clément Nouvel. "Among the players that have announced an FMCW Lidar for the coming years, none have achieved this so far." Small-scale production of such solutions could begin as early as 2026, but widespread adoption is expected to take another one to two years. The biggest challenges, according to Nouvel, include different chip substrate technologies, a relatively weak supply chain, and the lack of an automotive focus in photonics chip development.

Several developers currently see significant optimization potential in sensor placement within vehicles. For example, engineering service provider IAV offers simulation tools to determine the optimal sensor placement in vehicles, while two Fraunhofer Institutes are working on integrating radar, Lidar, and LEDs into the vehicle's headlight assembly to save space.

This article was first published at automotiveit.eu

Hands-free on the Autobahn in the BMW i7

As the first car manufacturer, BMW combines its Level 2+ hands-free function up to 135 km/h and the Personal Pilot at Level 3 in the 7 Series. Our practical test shows whether the systems deliver what they promise. The infotainment is also examined in detail.

Timo Gilgen