A Trip Through the Decades: LiDAR’s Journey from NASA to Today
LiDAR technology seems to be everywhere lately, in cars, phones, etc. However, it wasn't always like that. LiDAR had a long, 50-year journey to become the latest "hot" tech that it is today.
Over half a century following its inception, Light Detection and Ranging (LiDAR) has risen to prominence within multiple industries and technical applications.
A comparison of LiDAR, vision, and radar. Image used courtesy of Analog Devices
The technology, while suitable for unique purposes, has even been hailed as a successor to radar. Although LiDAR technology had been steadily progressing recently, moving from huge and bulky to more streamlined, it has a long history, but how did it become a staple technology for autonomous driving?
LiDAR’s Geological Debut
In the decades following World War II, the United States Geological Survey (USGS) focused on automating the surveying process. New technologies met aerial photography, and soon laser technology unlocked yet another efficiency level behind large-scale mapping.
Accordingly, LiDAR opened the door for rapid terrain imaging, which matured even more throughout the 70s. Remote, laser-based sensing enabled aircraft to map oceans, ice sheets, and forests.
Laser scanning was also pivotal in producing 3D images of lunar surfaces during the Apollo missions.
The LiDAR system was flown in the Apollo missions. Image used courtesy of NASA and James Abshire
Apollo 15 through 17 employed a form of LiDAR reliant on a “flashlamp-pumped ruby laser,” according to NASA. This method had a low pulse frequency of 3.75 per minute and relied on mechanical parts to function. The project was dubbed the Laser Altimeter Experiment.
Surprisingly, the LiDAR acronym originally existed as a blend between “light” and “radar.” The concept and name eventually morphed into the current form. It’s also known as Laser Imaging Detection and Ranging.
The first decade for LiDAR was mainly trying to find its footing, but once the 80s approached, LiDAR started to take shape fully.
The Advent of Next-Generation LiDAR
The 80s constituted a significant step forward for LiDAR technology mainly because of the arrival of solid-state technology, with motors and gears yielding to systems on chip (SoCs). That modern, compact packaging assumed control of its laser systems to scan a scene. Naturally, the advent of the semiconductor helped these LiDAR systems shrink in size.
Gains in compactness were coupled with the diode system's arrival. The improved reliability meant that NASA's LiDAR units could operate nonstop over multiple years.
The technology became instrumental in the following pursuits:
- Mapping the shape and surface of Mars, with over 600 million measurements taken,
- The initial expansion of Earth's topographical and atmospheric scanning, comprised of nearly 2 billion laser measurements, and
- Measuring the shape of various asteroids
When it came to diode LiDAR, it was markedly more efficient, had a longer lifespan, produced a denser laser, and boasted a much higher resolution. For example, measurements of Antarctica and Greenland were within a height resolution of just a few ems.
A diagram of NASA's MOLA. Image used courtesy of NASA
YAG lasers were renowned for their power and resulting penetration depth, making relatively light work of obstacles between the LiDAR source and target of interest.
Another technology that appeared during this time was dye lasers. Pulse-dye LiDAR operated at wider wavelengths and produced pulses as short as ten femtoseconds (or ten quadrillionths of a second).
LiDAR's growing ubiquity was seriously hampered before the 80s. Commercialization remained expensive, and a GPS shortage slowed the rollout of aerial systems. However, once producers scaled this hurdle, LiDAR was widely used for gathering geographical data and performing atmospheric research. Increased pulse frequencies and shorter laser wavelengths made measuring airborne particles much easier.
Each decade that passes, more and more advancements happen for LiDAR technology, so what did the 90s hold?
LiDAR Becomes Commercially Viable
Fast forward to the 90s, and LiDAR had started to gain steam with commercial buyers. These laser systems surged in popularity over their rivals. Since government agencies no longer had a monopoly on the technology, even though their funding and exploits paved the way for widespread adoption.
By combining onboard laser emittance scanning with aerial cameras, aircraft could easily capture 3D surveillance data while mapping. That data was extremely valuable from a surface analytics standpoint. While other mapping methods were inferential, sometimes making educated guesses to produce a map, LiDAR measurement was direct. Scans taken were much more reliable and rapid, leading to sensor-driven mapping becoming synonymous with less distortion.
Additionally, the prices of these units steadily decreased throughout the decade. LiDAR became more democratized as a result (yet still remained aspirational for many).
The technology also took a step forward as solid-state variants matured. While dye lasers were previously famed for their pulse speed, newer SoC-based lasers could produce pulses down to 5 femtoseconds. In one femtosecond, light only has enough time to travel about 300 nanometers. That's a distance slightly larger than the smallest bacteria on record. These laser wavelengths spanned from ultraviolet to infrared.
Overall, typical LiDAR systems contained either YAG, ruby, or YLF lasers until about 1995. Large receiver telescopes were also essential components. However, these LiDAR solutions were complicated, cumbersome, drew a large amount of power, and required the guidance of skilled operators. A change was needed if the technology were to grow in popularity.
Despite these challenges, LiDAR still had a lot of promise. For most of LiDAR's history, it has stuck to mainly aerial surveillance, so how did it move to vehicles like it is mainly known for today?
The LiDAR of Today
Though it has been a long, 50-year journey, LiDAR has come a long way. While some LiDAR systems can cost up to $75,000 apiece, costs have swiftly been falling fairly recently. It's now possible to acquire a commercial-grade LiDAR unit for just $500 via companies like Luminar. Additionally, it's predicted that LiDAR system prices could average out to roughly $700 each by 2025. That's incredible news for startups and customers alike.
Along with the wider availability arose a new application, with LiDAR no longer solely focusing on readings via rocket ship or satellite. The technology now lies much closer to home, as compact drones have become essential ground surveying aircraft, and autonomous cars are developing.
A possible roadmap for LiDAR in automotive and industrial applications. Image used courtesy of Yole Développement
It's now possible (and necessary) to procure a LiDAR unit measuring just inches in overall size. LiDAR is no longer relegated to massive "pucks" that span entire vehicle rooftops. Instead, automotive manufacturers can now acquire units fitted for car windshields.
Modern LiDAR is also much smarter, with each unit now capable of capturing upwards of 2 million data points per second with 5mm accuracy. Even compact units can send pulses accurately up to 1,000 meters away, a boon for autonomous safety.
Object recognition has also gotten better, and energy consumption has been slashed. LED and VCSEL diodes have shrunk sensor form factors considerably, to the point where smartphone camera housings can accommodate them.
Pulse modulation and single-photon avalanche diode (SPAD) systems have risen, which excel at detecting moving objects. They're also crosstalk resistant.
Accordingly, today's LiDAR systems have evolved to boast less latency and smaller bandwidth requirements. The software also plays as critical a role as ever. AI algorithms are fundamental in determining on-the-fly analysis of environmental objects.
Today's CMOS-backed sensors have unlocked real-time processing. They can also see in 1D, 2D, 3D, and 4D with RGB color. For the foreseeable future, companies in the automotive, IoT, farming, industrial, and forestry sectors (among others) will continue relying on LiDAR to tackle their biggest challenges.