Lidar Technology – How Does it Work and Where is it All Going?
A seminal, technical white paper, written by industry veteran and expert Brent Gelhar, providing an in-depth look into the physics of lidar technology from the beginning to where we might be going in the future.
I have been an amateur musician since I was in high school which was admittedly a long time ago. I decided on taking the path of engineering and corporate management when I recognized that to excel as a musician requires real inherent talent which I may not have had, and the engineering path I chose relied more on study and math skills (my sister, who is an accomplished pianist, along with many of her friends spent most of their careers struggling as musicians, I took a 30-year musical pause while the rest of my life and career got in the way and started playing again about 8 years ago).
This week my jazz band was having a practice and when we finished, our pianist and I struck up a conversation about laser scanning since he had read some of my recent articles (available on my website www.spatialinitiatives.com). He has been retired for several years now, but after having earned a Ph.D. in engineering, he spent almost his entire career in the nuclear power industry, researching and developing reactors.
He said “Boy, laser scanners sound like something the nuclear industry really could use”, which brought me to recall that one of the very first laser scanner manufacturing companies was Mensi (a French company established in 1987 and acquired by Trimble in 2003) which had worked very closely with Electricite de France (EDF) to solve specific problems relating to the decommissioning of nuclear power plants.
Another interesting connection with the nuclear industry is that the popular CloudCompare Open Source software for 3D point visualization and measurements which has been under continual development since 2004, originated jointly by EDF’s R&D and the Ecole Nationale Superieure des Telecomunications (ENST- Telecom, Paris). This work was the result primarily of Daniel Giradeau-Montaut and several other collaborators in their thesis work, and has brought free access to 3D data analysis to many without the financial means for the commercial products on the market today (please contribute here to support CloudCompare). Having looked back in time a bit, I decided that I should share some thoughts about the future of the technology.
It’s funny how jazz leads to lidar and open source software.
Current State – of – the – Art
For the past three or so decades, laser scanning devices have either been phase – shift based measurements of modulated laser light or time-of-flight (ToF). I will not bore you with a long technical explanation since there are many very good resources which describe this in great detail. Essentially, you take a laser beam, point it into an oscillating or rotating mirror, and measure the time difference from the laser beam leaving the mirror to the return of photons reflected back into the device.
In the most simple of distance measuring devices, the detector is placed separately beside the scanner, a binocular configuration. More sophisticated devices use a co-axial design, where the detector is in the direct line of the laser’s optical path, however this introduces more complexity (and of course cost) to the system while providing range and accuracy advantages. Knowing the angle that the mirror is pointing, one can map the distance and physical location of objects reflecting the light and build a 3D model of the environment.
I admit that this is a gross simplification of complex optics, electro mechanical and electronic systems, but it highlights how the underlying technology really has not changed much in this time. In fact, during the second World War, infrared light was being pointed at windows, the reflection being detected with a photomultiplier tube (the only photonic detector at that time) and the modulation of the photons being converted into sound, essentially eavesdropping (I know this as my father, who was an engineer developing radar systems in the German military explained it in great detail to me when I was quite young). These phase shift modulation decoding techniques are the essentially same as used in today’s phase – based laser scanners for measuring distance.
Timing- How Does it Measure so Quickly?
In 2012, I became involved with a novel timing technology which could precisely measure 2-3 picosecond events. Now to put it into perspective, we must understand that light travels at approximately 299,792,458 meters per second. This means that to travel 1 meter, a photon requires 3.3 nanoseconds. Since when using laser distance measurements for scanning we like to talk about millimeter accuracy, we need to be able to measure in the single digit picosecond times, such as a 1 millimeter distance is covered in 3.3 picoseconds, or 10−12, or 1/1,000,000,000,000 (one trillionth) of a second.
The timing technology had been developed in Latvia during Soviet times, and was used for decades (still in use today) for very precise measurements of the distance from earth – based telescopes to orbiting satellites with centimeter accuracy over 500+ km altitudes. It appeared that this technology may be a good fit for laser scanning so we raised €200k in EU development venture based capital funding and established a company in Latvia to explore building a laser scanner based on this timing technology.
Over a 3 – year period, we succeeded in building a bench laser ranging setup capable of measuring ranges out to 5 km with subcentimeter accuracy. Unfortunately, over the same period, new high speed timing chips and detectors became available on the market which, while not as high speed and accurate as our technology, they were adequate for the application, and cost in the sub $100 range versus our FPGA encoded discrete electronics which were in the thousands of dollars at production quantity.
The new integrated circuits were a major incremental change in the technology, so without being able to further raise funding we shuttered the company. The timing technology still exists and is currently being funded by European Space Agency projects for developing secure line-of-sight encrypted communications, delay measurements in optical fibers, quantum computing related among other applications. Details of the company and timing technology can be found at http://eventechsite.com/
With ToF devices, the actual timing of the pulses are typically achieved by measuring the voltage on an avalanche photodiode (APD) or photomultiplier (PMT) detector circuit and expressing that into time. We know when a laser pulse leaves the laser by triggering a capacitive circuit, when a pulse returns to our detector we stop the charging of the capacitor and read the voltage, translating that into time and therefore distance (since we know the speed of light as described above).
For phase-shift measurements, a continuous modulated wave of laser light is sent, we know the phase of the wave leaving the laser, that is compared with the phase of the returning light, so the distance between the waves correlates directly to the distance shift of the waves and tells us how far the light has travelled with greater accuracy than single pulse ToF systems, however with limitations on the range which can be achieved.
Now these descriptions are quite simplified since there are many optical and precision electronic circuit considerations to be designed into the laser ranging system, however the underlying theory is as simple as this.
Detectors- How We Actually See the Light and What’s Next
The very first photodetectors were vacuum tube photomultiplier tubes, which are still in wide use today for applications ranging from atomic physics to biomedical microscopy. With the advent of semiconductor electronics in the 1940’s and 1950’s new photosensitive diodes emerged, one being the APD, which is probably the most widely used high speed detector for lidar applications. About 10 years ago, methods were developed where arrays of APD’s could be arranged on a chip resulting in a multiple point detector, think of it as a camera sensor but for measuring the time-of-flight of photons.
Advanced Scientific Concepts Inc., Santa Barbara CA, (ASC) developed a series of conventional APD arrays, capable of real-time ranging video at some tens of meters (Continental Corp, Auburn Hills MI, acquired the ASC automotive products in 2016 for their autonomous vehicle product developments).
Another technology, based on InGaAs photon counting arrays sensitive in the mid-IR range (approx. 1500 nm wavelengths which is very advantageous for lidar imaging applications), developed by MIT’s Lincoln Labs and commercialized by the Princeton Lightwave company (acquired earlier this year by Ford’s Argo AI, LLC, Pittsburgh PA) has progressed to a point where the detector arrays had reached 256×256 elements and more, with the capability of measuring ranges out as far as several kilometers with very high accuracy. All of these technologies created the first Flash Lidar cameras, meaning being able to “flash” a laser pulse and measure all the array image points simultaneously, as opposed to conventional single point detectors which need the laser beam to be serially scanned on each measured point.
More recently a new generation of PMT’s have emerged from SensL Corp, Cork Ireland (now part of ON Semiconductor, Phoenix AZ) which are solid state PMT arrays, manufactured on silicon semiconductor wafer technology. These silicon photomultiplier arrays (SiPMs) are extremely cost effective devices, operating in the visible and near IR wavelength ranges having mostly been deployed in biophotonics and physics related applications.
So what we have seen in the past four decades are advances in photodetectors coming on the market, but followed with only incremental improvements to the very high speed versions.
Scanning Technologies- Mechanical Versus Solid State
This area is becoming very exciting now. The advent of ultra-high density neodymium magnets has revolutionized everything from hi-fi loudspeakers to DC motors. In the past, DC motors all required heavy magnets, coils and carbon brushes which generated much electrical noise. This new generation of magnets allowed for the development of galvanometers, where precise rotational position accuracy can be achieved, and compact brushless DC motors. Without this advance in magnet and motor technology, the current drone industry revolution would never have happened without the miniaturization and weight efficiency gains.
The same technology has grown a whole family of linear actuators, such as voice coil mirror actuators.
In the early 1990’s we saw micromirror devices emerging commercially. The most popular of these is the Digital Light Processor from Texas Instruments which are employed in movie and presentation projectors. Micromirror technology has been employed in many medical diagnostic devices also. A major drawback of this technology in lidar is that the deflection angles are typically limited to narrower angles, and one must remember that they are still mechanical devices which suffer from hysteresis distortions from the movement which often degrades the laser beam profile.
Most exciting are more recent developments of true solid state beam deflection devices. These range from electro optical scanners (EO) which employ a thin liquid crystal layer to control beam deflection, and acousto optical scanners (AO) where a piezoelectric transducer stresses a crystal (usually quartz) causing light passing through the crystal to deflect at a known angle.
The truly promising devices for the lidar technology industry are semiconductor chip – based optical phased arrays (OPA) and optical deflection waveguides (ODW). Both of these technologies are now leaving the laboratories and have the benefits of highly controlled deflection angles, high speed, reasonable power throughput and low power consumption. This combined with no moving parts are some of the most desired attributes for laser beam pointing application, and especially in lidar technology.
I have assembled a summary of optical scanning and technology as it relates to the lidar industry (see infographic figure 1) and some of the attributes of the different methods.
What we can see as the trend that is driving all these advances are mass market adoption of lidar technology into the automotive industry. This is helping to drive the costs of lidar devices and components down to a point where they are easily deployable in very wide transportation applications. In turn, the components are now much more accessible to smaller niche applications, such as remote sensing and surveying applications.
This is the same revolution we experienced with GPS technology in the 1990’s. When small handheld consumer and automotive navigation applications started embracing the expensive positioning receivers, quickly the costs and form factors of GPS devices came within mass market adoption. Just look at how today almost every cellphone has GPS embedded in it. We are even seeing the first cellphones with 3D sensors now emerging on the market so mass adoption of 3D measurements and imaging applications is clearly on the way.
Note – If you liked this post click here to stay informed of all of the 3D laser scanning, geomatics, UAS, autonomous vehicle and Lidar News. If you have an informative 3D video that you would like us to promote, please forward to firstname.lastname@example.org.
lidar technology, lidar technology, lidar technology, lidar technology