Imagine this: a car traveling at night along a winding country road. No headlights. No hands on the steering wheel.
A recipe for disaster?
In the future, it might not be, with Light Detection and Ranging systems (lidar) guiding the way.
The automotive industry is devoting massive amounts of resources to the development of autonomous driving. The key to propelling this technology is the use of high-quality components, particularly those made with specialty glass, because it can be engineered to offer unique mix of properties, like thermal resistance, strength, impact resistance, and optical transmission, that other materials cannot.
Lidar versus camera-based
There have been giant leaps forward in the last twelve months, but the industry has a long way to go before fully autonomous driving hits the market.
Broadly speaking, there are already two types of autonomous driving system on the road. Toyota, for example, uses a radar-based system that will automatically reduce a vehicle’s speed if it gets too close to a car in front of it. And then there are camera-based systems. These can interpret images that, with a bit of machine learning, can identify cyclists, or pedestrians, and may be able to “read” signs.
But both radar-based and camera-based systems tend to have drawbacks that lidar overcomes. A radar-based system, for example, can tell how far away an object is, but it cannot tell what the object is. Camera-based systems have some difficulty judging distance and the camera-based systems have difficulty making sense of the world. The broad panel of a truck can be mistaken for open sky, or telling the difference between a puddle and a patch of thin ice.
There are benefits to both systems - radar-based driver assistance is quite handy, and camera systems are helping to solve some of the machine learning problems presented by driverless cars. The fully autonomous vehicles of the future will likely include these technologies, which will likely blend many different technologies to create redundancies in case of failure. The way in which lidar overcomes some of the drawbacks of radar-based and camera-based systems means that it will have a prized place in a robust systems that will likely feature multiple redundancies.
Using laser light beams to scan the surrounding environments, lidar can judge both the distance of objects, and to create accurate 3D point clouds that can be processed to understand the visual environment. By taking the best of both worlds, lidar, together with radar and cameras linked to artificial intelligence systems, will offer the surest path to fully autonomous driving.
There are now more than 100 companies, from boot-strapped startups to major OEM automakers, working on lidar in some form. But because lidar research is still in its infancy, there is no consensus on what these products look like or how they will solve certain problems. In general, however, lidar units for automotive applications tend to share a few common features.
Demand for strength and thermal stability
The part of the unit that will be most visible to consumers will be the protective window. It’s the transparent material that protects other components from damage. Here, strength and thermal stability are important properties. These units must be able to withstand rocks, tree branches, rain, and hail without breaking. They may find themselves subject to rapid temperature fluctuations (say, from driving up a large mountain) that would cause some materials to crack or warp. At the same time, these materials need to ensure superior light transmission of near-infrared light, while mitigating potential interference from visible ambient light.
Specialty glass fits this bill, especially when compared to materials like polymers that many assume appropriate for these applications. Certain glasses can be engineered to have a very strong microstructure that results in high material hardness, and they don’t “yellow” in response to long-term sun exposure. Coatings can be applied to increase the hardness of the material, and anti-reflective coatings can increase light transmission. All of these properties combine to ensure long-lasting, reliable operation of lidar systems.
Glass also plays an important role in the optical path. When light enters the lidar unit, it is generally directed via a series of mirrors and beamsplitters to a diode or sensor at the base. Along the way, it may pass through filters to help cut down on signal-to-noise ratio.
Many lidar systems for automotive applications typically use lasers diodes at two frequencies: 905 nm or 1550 nm. Companies have chosen these diodes because they are readily available, and because they operate within the near-infrared range, above the visible light spectrum. This is in contrast to aerial lidar systems, which tend to use lower frequencies which span the visible section. At altitude, there is reduced risk that lidar signals could impact the human environment.
Clear transmission required
The decision over whether to rely on one diode or the other has much to do with the design philosophy of the company making a device. Diodes on the 1550 nm wavelength tend to resist absorption from water. They also use less power, and consequently, a longer range. But they tend to be slightly more expensive than 905 nm diodes. For the purposes of transmission through filters and substrates the composition of substrate material is very important. High-quality glass materials, made with very pure ingredients such as Schott’s Borofloat can achieve transmission rates of 92%. D263 T eco glass as a lidar filter substrate provides very high transparency across a wide range. It is also suited for anodic bonding and chemical toughening.
The quality of glass is key with optical lenses as well, because of the high performance and reliability requirements that lidar systems will demand. Lenses must deliver good, long-lasting image quality regardless of temperature differences or climate conditions. In some cases, weight and size will be important to achieve design and miniaturization goals. Designers of lidar systems are exploring athermic lenses with high refractive indexes to meet some of these demands, and aspheric lenses for cases where compactness is a priority.
A final class of components is critical for the operation of any electronic device that operates in harsh conditions but is little talked about. Glass-to-metal hermetic sealing protects electronics from condensation, while enabling optical and thermal performance in safety critical environments.
A can’t-fail system, with time to grow
In many parts of our lives, a 99% success rate is admirable. With autonomous driving, that statistic would be disastrous. High-quality components, particularly glass, should form the basis for lidar designs, because it is the only material that can offer the unique combination of properties that these systems demand.
Luckily, there may be a lot of time for companies and engineers to get things right. There are some fundamental questions about power output, range, the number of pixels an image should have, and the amount of energy and computing power your car will need. In many cases, the answers to these questions, like the choice between laser diodes, will involve trade-offs between cost, performance, and a host of other factors.
Designers of these systems would be wise to work with experts in material science early in their processes to provide a roadmap to success.
Written by Michael Hardbarger, Business Development & Sales Manager at SCHOTT.