By John Oncea, Editor
Mercedes-Benz plans a limited launch of its Drive Pilot system in 2024, the first to market with a Level 3 autonomous driving system. LiDAR is playing an instrumental role in the progression to Level 3 autonomy, aided by the unsung hero: LiDAR filters.
When Johnny Nash released his number one hit “I Can See Clearly Now” in 1972 he was, of course, singing about LiDAR filters.* And for good reason – the automotive industry’s use of LiDAR filters is enabling cars to sense and respond to the surrounding environment with greater precision, something that will help the industry move to vehicle autonomy to Level 3.
We’ve already established that LiDAR is a pretty great technology that is expected to improve in terms of resolution, range, and accuracy in both the short and long terms. These improvements will make LiDAR even more effective at detecting and navigating complex environments.
Let’s take a look at the LiDAR filter’s contribution to allowing you to, one day, take a nap while driving cross country.
* No, he wasn’t.
No To Smoke, But Yes To Mirrors
In the simplest of terms, a LiDAR system involves a laser range finder reflected by a rotating mirror. The laser is scanned around the scene being digitized, in one or two dimensions, gathering distance measurements at specified angle intervals. A complete LiDAR system is made up of several components working together to generate, record, and georeference the data. These include a LiDAR source, detector, and scanning mechanism; timing electronics; a global positioning system (GPS); an inertia measurement unit (IMU); and a computer.
An often-overlooked component of a LiDAR system is filters, something LiDAR sensors use to isolate target signals and prevent other light from reaching the detector. These filters need to be able to maximize signal-to-noise ratios, consider the sensor platform and environmental conditions, and isolate the target LiDAR return signal. Consideration needs to be made to factors such as temperature range, operating range, and wavelength when discussing LiDAR filters, of which there are several types including near-infrared filters, short-wave infrared filters, and ultra-narrowband interference filters.
The specific type and combination of filters used in a LiDAR system depend on the intended application and environmental conditions noted earlier. LiDAR technology continues to evolve, leading to the development of more advanced and specialized filters to meet the demands of various industries, including autonomous vehicles. Here are some common types of LiDAR filters:
- Range filters: These filters remove data points that fall outside a specified range. They help eliminate outliers or objects that are too close or too far away from the LiDAR sensor, which can distort the accuracy of the resulting point cloud.
- Intensity filters: Intensity filters are used to refine LiDAR data based on the intensity of the reflected laser pulses. They can be used to enhance the quality of the captured data by removing points with low intensity, which might correspond to weak reflections or noise.
- Temporal filters: LiDAR systems often employ temporal filtering to account for changes over time. This helps in distinguishing between static objects and moving objects by analyzing multiple scans taken at different times and filtering out dynamic elements.
- Spectral filters: Some LiDAR systems use spectral filtering to differentiate between different types of objects based on the wavelengths of the reflected light. This can be useful in applications such as vegetation classification or material identification.
- Noise filters: These filters are designed to reduce noise caused by various factors such as environmental conditions (like rain, fog, or dust), sensor imperfections, or interference from other sources. Filtering out this noise improves the accuracy of the captured data.
Many LiDAR sensors are mounted on satellites, airplanes, UAVs, and autonomous vehicles – all of which require the sensor to function under harsh environmental conditions with little to no maintenance. Because of this, thin-film interference filters are commonly employed because of their inherent durability and lack of a need for maintenance or calibration.
Seeing Clearly With A Little Help From Our Friends
Our friends at Alluxa, Inc. note it is the precise nature of LiDAR return signals that makes the use of ultra-narrowband thin-film interference filters necessary, writing, “These filters must be able to achieve high transmission over an ultra-narrow bandwidth to isolate the return signal, and deep out-of-band blocking over a large wavelength range to attenuate sunlight and other extraneous light.”
There exists, however, a variety of LiDAR systems, each requiring specific filter requirements to enhance signal-to-noise ratios. For example, laser altimeters need ultra-narrowband interference filters of less than 1.5 nm at full-width half maximum (FHWM), with over 90% transmission at the laser wavelength. They also require blocking of greater than OD6 (-60 dB or 0.0001% transmission) for out-of-band signals between ~300 – 1300 nm. However, Raman LiDAR filters must have steep edges to allow the Raman signal to pass through the detector while blocking stronger elastic backscatter signals at the laser wavelength to a level of OD8 (-80 dB or 0.000001% transmission).
“LiDAR systems also require that the filters’ thin-film coating must be as uniform as possible,” Alluxa writes. “When uniformity is not controlled, the thin-film layer thicknesses vary across the surface of the filter, resulting in a location-dependent wavelength shift of the filter spectrum across the clear aperture.
“If a filter with uncontrolled uniformity is integrated into a LiDAR system, a large number of LiDAR return signals will end up being blocked by the filter and will not reach the detector. Fortunately, a uniformity-controlled thin-film coating will ensure that target signals will not be blocked by the filter.”
Level 3 Autonomy And Beyond
As we approach Level 3 autonomy, the significance of LiDAR technology in ensuring active safety cannot be overstated. While cameras are undoubtedly valuable for infotainment and some aspects of vehicle perception, LiDAR fills the gap when it comes to critical active safety features. The benefits of LiDAR's accurate depth and motion sensing are not only crucial for achieving Level 3 automation but are also fundamental to attaining higher levels of autonomy.
“If detectors are the ‘eyes’ of a LiDAR system then optical filters are the ‘sunglasses’ – reducing glare to allow the eyes to see what they are looking for without the background noise,” writes another friend of ours, Iridian Spectral Technologies.
As LiDAR systems become more common in the automotive industry, there will be a significant demand for low-cost and high-production capacity filter supply chains. It is not enough to simply find a technical solution; it is crucial to develop a scalable solution that can meet the rapidly growing need for affordable sunglasses for all the autonomous vehicles that will soon be on the roads.
“There is still a long road ahead before the streets are filled with Level 3 autonomous vehicles, but we have begun down the path and are already reaping the benefits of automotive photonics sensor technologies and the optical filters that enable them to see clearly,” Iridian writes.
In other words, we are beginning to see clearly.