Summary: This story explores five key challenges facing today’s LiDAR systems—and how they are being overcome. It also demonstrates the broad application potential for LiDAR once these obstacles have been removed.
LiDAR technology has gained considerable ground as a viable approach for many applications—interiors and exteriors of automobiles, autonomous vehicles, and drones, to name just a few. However, five challenges stand in the way of advancing conventional LiDAR architectures for use on a mass scale as a cost-effective solution that improves automobile performance and safety, addresses more industrial processes and applications, and adds new functionality and capabilities to consumer products.
The LiDARs being used today in driving trials are very useful for training the AI systems, but will not be deployable in production vehicles due to not meeting size, reliability, and cost requirements. There are new LiDAR systems in development that can meet all the requirements for broad production deployment, but first, let’s review the challenges to overcome.
The Five Major Challenges Facing Today’s LiDAR Systems
These five LiDAR challenges are significant and until they are resolved, we’re only realizing a fraction of LiDAR’s potential as a broadly deployable technology in production vehicles.
Many of the existing LiDAR systems are bulky rotating units highly visible on the roof of a car. This technology does not meet the size, cost or automotive qualification requirements for OEM mass market deployment. While there are some smaller implementations of rotating LiDAR units that broaden applications in automobiles by allowing multiple units to be embedded into various areas of the car, these units are still too big, or have traded size for performance (resolution and range). Continuing to compress complex rotating machinery into smaller packages or using technologies and components that cannot be further scaled in cost or meet automotive Grade 1 qualification requirements will yield diminishing returns, calling for a fundamental change in technology.
LiDAR demonstrations often generate enthusiasm in customers who unfortunately find later that the winning features of demo models depend on expensive, over-specified components that will not be able to meet automotive grade qualification. The major cost drivers have been alignment of the lasers to the optics, achieving enough output power to meet range requirements, and employing a method to have laser beams cover the required field of view.
Many of the technologies used in LiDARs deployed in trials today will not be able to be reduced in cost for mass market adoption. Despite claims that increased production will drive down costs, higher volumes and design improvements won’t be enough without significant design revisions.
Most of today’s LiDAR systems are mechanical, or Micro-Electro-Mechanical Systems (MEMS) based, with all the disadvantages of moving parts, including limited reliability. These products require careful calibration and are expensive to manufacture. Because most aren’t qualified to operate over the automotive Grade 1 temperature range (-40°C to +125°C), they have high failure rates under thermal stress. The clear answer is to replace the moving parts with a solid-state alternative with each component able to meet Grade 1 temperature and qualification. However, even solid-state technologies such as Lasers have a way to go — there are very few lasers today that meet Grade 1 automotive requirements.
Contemporary LiDAR systems have demonstrated operation over relatively long distances, but a widely-held assumption is that this level of performance is only possible with 1,550-nm scanning lasers. This technology does deliver excellent range, but at a very high price, due to both the high power 1550-nm lasers as well as the sensors. Inexpensive flash laser technologies with comparable range and attractive cost already exist.
5. Eye Safety
Typically, 1,550-nanometer light has long been considered the sweet spot for eye safety. It is indeed a safer solution than other laser types that penetrate they eye to the retina. However, the dominance of high-cost 1,500-nm DPSS, diode pumped fiver lasers or DFB with EDFA lasers is being challenged by new device topologies that make shorter wavelengths safe.
Tackling the Size Challenge
What’s the good news? LiDAR’s days of bulky spinners on a rotating gimbal atop vehicles are numbered. A LiDAR system’s footprint can be greatly reduced by using the right Vertical Cavity Surface Emitting Laser (VCSEL) illuminators. Illumination solution manufacturers can compress hundreds of VCSEL lasers onto a microchip that’s mere millimeters in size.
This technology brings the potential to greatly reduce the entire footprint of the LiDAR system while still delivering the eye-safe high power required to illuminate an entire field of view. The LiDAR unit can then be reduced to roughly the size of a deck of cards, or smaller. In automobile applications, this allows LiDAR to be placed in multiple areas on the exterior and interior of a car to effectively gauge distances of objects that may be only a few, or as much as 200-plus meters, away. These changes offer added environmental monitoring and crash avoidance capabilities for autonomous vehicles.
Smashing Cost Barriers
The same technology and thinking behind the size solution can also drive cost savings as well — drastically! Semiconductor Vertical Cavity Surface Emitting Lasers (VCSEL), specifically those with 940-nm wavelengths, exploit existing semiconductor processes optimized for mass-produced products like cellphones, using silicon chips instead of the far more expensive Indium Phosphide (InP) and Indium Gallium Arsenide (InGaAs) devices found in 1,550-nm lasers and sensors.
When you combine these process and materials savings, each LiDAR unit can achieve cost points of $200 or less, as opposed to the $1,000-plus costs for many of today’s mechanical and 1,550-nm solutions. At these prices, using multiple LiDAR units on autonomous automobiles for advanced sensing capabilities and increased safety is much more cost effective.
Most of today’s LiDAR technologies are MEMS or rotating mirror based. These technologies include many moving parts, they are expensive to produce and have high failure rates. It’s also important to note that many LiDAR systems must operate in less than ideal climates and high-vibration environments. The right 940-nm VCSEL modular laser technology is designed to operate at the automotive AEC-Q100 Grade 1 temperature ranges from -40°C to +125°C. This technology has been deployed in the telecommunications arena for more than two decades, with very high reliability. In addition, these new VCSELs have been tested for millions of hours of equivalent operation at the power levels for long range LiDAR, and at AEC-Q100 Grade 1 temperature ranges without any unexpected reduction in performance.
Typically, 905-nm Edge Emitter Lasers (EELs) are limited to around 100m of range due to eye-safety limits while 1,550-nm lasers can be used to achieve long range, eye safe illumination. However, the cost of lasers and 1550-nm sensors is prohibitive for the foreseeable future. The LiDAR industry has created its own myth by thinking the only way to achieve this range is by using these expensive 1,550-nm lasers and sensors.
The 940-nm lasers offer a unique combination of exploiting the dip in the solar spectrum at 940 nm (less solar interference) and leveraging high volume, low cost VCSEL technology. The reduction in ambient light noise at 940 nm enables better range at lower laser optical powers. You may have already used 940-nm VCSELs in a mobile phone for 3D sensing at very low powers and short ranges. Now, VCSEL technology can achieve very long range and be eye safe. Read on to find out how this is being done.
Innovating for Eye Safety
Due to eye safety limitations, 905-nm Edge Emitter Lasers (EELs) are typically limited to around 100m of range. Although high-power continuous wave near infra-red lasers are unsafe for eyes because they can penetrate to the retina and can damage vision, high power 940-nm sources that are pulsed for a very short duration (billionths of a second pulses) at lower duty cycles can be tolerated by the eye (“eye-safe”). Today’s 940-nm VCSEL arrays are entirely different from a high brightness point source such as an edge-emitting laser (EEL).
The 940-nm VCSEL arrays contain individual VCSEL lasers that are each relatively low power, use short pulses and are therefore eye safe. This technology operates using narrow pulses (billionths of a second) at very low repetition rates. When these VCSELs are combined into an array, it creates a distributed source with very high output power to achieve the long range, but at a very low average power and with distributed point sources which do not damage the retina. As an example, TriLumina has developed eye-safe, flip-chip 940-nm pulsed VCSEL arrays with 600W peak optical power enabling over 250m range, with low duty cycles where average optical power is only half a watt.
Widespread LiDAR Deployment Is Possible
While today’s MEMS and mechanical LiDAR challenges are indeed obstacles to broader deployment, they can be overcome by promising new VCSEL laser technologies that are completing automotive qualification in early 2019. Imagine autonomous vehicles with sensing capabilities, both on the inside and outside of the car, that dramatically reduce the more than 37,000 annual traffic fatalities and allow for autonomous driving to be realized.
These goals can only be achieved by reigning in costs, reducing size, and meeting automotive reliability requirements and range, all while ensuring eye safety. With the right laser technology, LiDAR can take the next steps toward broader deployment.
At TriLumina, we are developing VCSEL illumination products for the automotive ADAS industry and tackled each specific area of performance, size, cost, eye-safety, and automotive Grade 1 qualification to enable broad deployment of LiDAR. This is based on patented flip-chip, back-emitting VCSEL arrays that combine high pulsed power arrays, integrated micro-optics, and electronic beam steering on a chip.
About the author
Brian Wong is President and CEO of TriLumina Corp. He has more than 35 years of experience in technology and has held the title of President and CEO at four technology companies.