LiDAR is rapidly gaining interest and being deployed in ADAS and automated vehicle sensing systems, but there are different ways to implement the technology. This paper explains these approaches and the relative advantages of coherent LiDAR detection.
Light detection and ranging (lidar) was first conceptualized in the 1930s, almost simultaneously with radio detection and ranging (radar). However, the technology was not proven until the advent of lasers in the 1960s, and in the following years, the development of optical communications led to significant advances in lasers and optical modulation technology.
In 2008, the first commercially available LiDAR system, initially called 'optical radar', made its debut in a Volvo passenger car. This breakthrough technology powers one of the first Automatic Emergency Braking (AEB) systems, enabling vehicles to brake automatically to prevent or mitigate rear-end collisions.
Following its early introduction 15 years ago (and its subsequent replacement by radar as a cheaper alternative to AEB), high-resolution lidar has rapidly evolved to become a key high-resolution sensor for self-driving car programs and has fostered a variety of technologically innovative and well-funded start-ups. Offering greater range, superior resolution, and real-time 3D visualization of the vehicle's surroundings, the technology is now maturing into an important sensor paradigm not only for autonomous driving, but also to complement Advanced Driver Assistance Systems (ADAS) in passenger cars and commercial fleets.
LiDAR sensors emit photons in the infrared spectrum to detect and create 3D images of their surroundings. They are proving to be very popular in automotive applications. The main advantage of LiDAR over radar is that the light used has a very short wavelength, which allows for precise measurements. In addition, LIDAR can operate in any lighting conditions and has a better detection range compared to cameras. The data captured by LIDAR sensors can be regarded as a "point cloud".
There are many things to consider when developing a LiDAR system, such as what wavelength to use, how to scan, and how to deal with interference. However, the biggest decision the system has to make is how best to detect the returning photons. There are two main contenders, direct detection and coherent detection.
Direct Detection
In a direct detection system, a laser pulse is emitted, effectively starting a timer. When the return of the laser pulse is received, it stops and calculates the distance based on the elapsed time. See Figure 1.
Figure 1: Since the speed of light (c) is constant, the distance to the target is Δtc/2, where Δt is the time between the start of photon transmission and the photon reception front.
For distances up to about 50m, there is no need for a high-quality tunable single-mode laser (since it is simply a source that compresses a large number of photons in a short time) or modulation, thus simplifying the drive circuitry. Nor is precision optics required to compensate for wavefront aberrations.
Why the short range? As the illumination area increases with distance, the return power decreases (as the square of the distance). Note: The formula for calculating this is: return power is approximately equal to transmit power x (target area/illumination area) x (receive area/(π x range2)). This loss cannot be avoided, so the simplest solution is to transmit more power or increase the receiver sensitivity.
However, there is a limit to the amount of laser power that can be used. Intense near-infrared (IR) light (800 to 1400ηm) can impair human vision. Simply increasing the transmission power of NIR light in ADAS or self-driving car applications can pose a danger to other road users and pedestrians.
To improve the receiving sensitivity, photon collection can be increased by using larger area receiving lenses. In addition, avalanche photodiodes (APDs, photodiodes with intrinsic gain) can be used, although they tend to be expensive, fragile and small (which further complicates the system optics) and can only provide a gain of about 15 times before self-generated noise becomes a A problem. Other sensor types, such as Geiger Mode Avalanche Photodetectors (GMAPDs) and Single Photon Avalanche Detectors (SPADs), provide better sensitivity in direct-detection LIDAR systems, but are less effective in snowy, dusty, or foggy environments.
In addition, all detection systems require some form of interference mitigation. Whether it is radar or LIDAR, the system needs to know that the signal (either pulsed radio waves or photons) being received by the receiver is coming from the transmitter. Interference issues arose in the early days of pulsed automotive radar. Once many cars were equipped with radar, mutual interference became a problem. The most popular solution was to switch to coherent detection techniques, where the radar system primarily uses frequency modulated continuous waves (FMCW - see below).
Another limitation of direct detection lidar is that it does not directly measure the velocity at each point - instead, it must be calculated by determining how the range changes over time (i.e. comparing multiple subsequent frames), which can impair system responsiveness.
Coherent Detection and FMCW
This involves mixing samples of incident light with transmitted light, which has two main benefits. First, noise-free amplification of photon gain can be achieved through phase-length interference (i.e., the received signal is multiplied by the transmitted signal), which allows the system to achieve excellent sensitivity with very low power lasers. Second, mixing the transmitted and received signals makes the LiDAR system highly selective, as light that is not exactly the same wavelength (e.g., sunlight or light from neighboring LiDAR systems) is simply ignored.
There are several ways to implement coherent detection LiDAR systems, but the most popular is frequency modulated continuous wave (FMCW) modulation. Figure 2 shows a simplified example.
Figure 2: The laser operates around 1550 nm and modulates at several hundred MHz (e.g., from 1550.002 to 1550 nm). The emitted (and reflected) signal is about 200 THz. After optical mixing, the photodiode presents the sum and difference of the two signals. The photodiode has a limited bandwidth and does not respond to sums of about 400 THz and can only detect difference signals of a few hundred MHz.
In practice, the laser is scanned up and down in frequency to produce a sawtooth profile (frequency vs. time) from which distance and velocity can be derived; for the latter, consider the Doppler effect. Figure 3 shows a more detailed overview of the optics.
Figure 3: Main optical components of an FMCW lidar system.
Although more complex than direct detection systems, FMCW lidars have many advantages. For example, as mentioned above, the return signal is multiplied by the sample acquired from the transmitting source (local oscillator, LO in Figure 4). Due to the high path loss of the lidar, even a few percent of the LO can be much larger than the return signal. The amount of signal amplification is very high, but limited to signals of exactly the same wavelength, which leads to high photon efficiencies.
For example, an FMCW lidar system with a range of about 300m can be realized with a laser power of less than 200mW. For the same range, a similar direct detection system would require 1000 times the peak power. Notably, FMCW is at the heart of other areas of LiDAR; for example, optical altimetry instruments with ranges of up to several kilometers and laser Doppler LiDAR instruments for wind characterization with ranges of more than 500m.
Another advantage of coherent lidar is the rather low bandwidth of the signal chain. If we consider the wavelengths in Fig. 3 (where the laser scans from 1550.002 to 1550ηm), the photodiode bandwidth can be limited to a few hundred MHz. direct detection systems require as wide a bandwidth as possible (typically more than 2GHz) in order to resolve the leading edge of the received pulse.
Understandably, the narrower bandwidth allows the use of lower noise mutual impedance amplifiers and slower analog-to-digital converters on the photodiode.
Finally, coherent detection provides per-point velocity information. The benefit of per-point velocity is that it is an additional contextual metric that subsequent sensing systems can use when interpreting LiDAR (and other sensor) data, making it possible to make more informed decisions.
The various benefits of coherent detection are therefore significant, but coherent lidar is not without its challenges.
The laser must be able to maintain its phase integrity over a long enough period of time for its light to reach and return from the farthest target. If the phase of the laser changes significantly over the transmission time, coherence may be lost and may result in blurred distance measurements. In addition, it must be FM (in the case of FMCW). Most diode lasers are not up to the task, but some semiconductor tunable lasers have appeared on the commercial market.
In addition, not all scanning mechanisms are compatible with coherent detection. The receiver needs to observe each point long enough to allow the light to reach and return from the farthest possible target, since the return signal needs to be mixed with the transmitted signal. For example, a range of 300 m requires the scanning mechanism to remain stationary for at least 2 μs, but many continuously moving scanning mechanisms are unable to do so.
Finally, it is important to note that the signal processing task of coherent lidar is significantly greater than direct detection. Fortunately, semiconductor manufacturers have introduced powerful system-on-chip (SoC) products that integrate data converters, microcontrollers, and DSPs with FFT gas pedals to meet these signal processing needs: indie Semiconductor's iND83301 Surya LIDAR SoC is one such example.
OVERVIEW
Different lidar applications benefit from different design approaches. As mentioned earlier, high-power pulsed direct detection can work well in applications such as airborne ground surveys where ultra-long ranges are required and where there is little risk of LiDAR systems interfering with each other or harming the human eye.
However, for applications such as ADAS and automated ground vehicles that require a range of <1km and where other potentially interfering LiDAR systems are likely to be deployed, coherent detection (and in particular FMCW) offers a number of advantages. These include immunity to interference (including solar), high signal-to-noise ratio (important in adverse weather conditions), locally accurate velocity detection (providing additional information to the sensing system), and ease of system modification. For these reasons, coherent LiDAR detection is gaining momentum given the multiple use cases, especially next-generation automotive sensing.
