New LiDAR Chips Go the Distance, Rain or Shine
Though LiDAR is one of the most popular technologies for ADAS, it still has its hangups. Now, developers are fine-tuning LiDAR at the chip level—for instance, by extending detection distance and priming it for inclement weather.
LiDAR remains one of the top technologies garnering attention for ADAS vehicles. Forbes contributor Sabbir Rangwala comments, "LiDAR has occupied a “must have” status by a majority of automotive OEMs for ADAS, and AV driving stack companies for localization, mapping, and Level 4 autonomous driving."
Block diagram of a LiDAR system. Image used courtesy of Marco Van Uffelen
With such prominence, how are engineers and developers innovating this technology at the chip level? In recent months, many companies have zeroed in on extending LiDAR's detection distances, its functionality in hazardous weather, and its system integration.
LiDAR System Detects Objects Over 200 Meters Away
NeoPhotonics is one of the leading developers in silicon photonics and high-level IC-based lasers, modules, and subsystems for intense bandwidth applications. With the goal of improving road safety for autonomous vehicles, NeoPhotonics is increasing ADAS reaction time at longer distances through something called "coherent LiDAR technology."
Coherent LiDAR obtains a target range by measuring the frequency difference between a local oscillator and return signal. This process yields high-fidelity signal detection. By using Doppler shift to determine the phase of the return signal, coherent LiDAR simultaneously measures range and velocity.
Recently, NeoPhotonics announced a frequency-modulated continuous-wave (FMCW) laser module as well as high-power semiconductor optical amplifier (SOA) chips. These solutions are said to enable long-range automotive LiDAR for ADAS, improve high-resolution sensing for industrial applications, and enhance data acquisition and transmission in telecommunication systems.
Diagram of a coherent LiDAR transceiver. Image used courtesy of NeoPhotonics
NeoPhotonics claims the controllable and tunable high-power FMCW lasers will enhance existing LiDAR systems’ ability to detect objects at a further distance; this is thanks to a narrow linewidth of 1550nm. The 1550nm wavelength is what NeoPhotonics calls ”eye-safe” compared to other lasers.
The SOA chips implement coherent LiDAR devices such as PIN receivers. These receivers are a type of photodiode used to convert optical signals into electrical signals and can achieve 200- to 400-meter detection while mitigating external light interference.
NeoPhotonics asserts that devices with coherent LiDAR technology are more accurate in measuring the distance, which may offer a key advantage over traditional electromagnetic LiDAR.
LiDAR Takes on Inclement Weather
How do LiDAR sensors and lasers respond when inclement weather is thrown into the mix? This is a question Amazon's autonomous vehicle company Zoox is exploring on the rainy streets of Seattle. Because LiDAR sensors historically struggle with object detection when blocked by ice, snow, fog, or downpours of rain, Zoox hopes testing in the Seattle area will help the company solve these issues through trial and error—and it will subsequently update its hardware and software.
Zoox thinks Seattle will be the perfect location to test weather-resistant LiDAR. Image used courtesy of The Verge
Likewise, Tower Semiconductor has recently turned to CMOS image sensing technologies to improve autonomous systems in poor weather conditions. Similar to NeoPhotonics, Tower Semiconductor is operating at a 1550nm wavelength with FMCW. Tower Semiconductor, however, is also incorporating 3D vision.
Tower Semiconductor's LiDAR systems include CMOS image sensors, which offer visibility enhancement at low-light conditions. Additionally, the company includes a camera with a high-dynamic range of resolution. Coupling the camera with a CMOS sensor will allow LiDAR systems to create a 3D image of the surrounding area.
Resolution difference between a regular high-resolution sensor (left) and a gated sensor (like Tower Semiconductor's CMOS image sensor—right) for high dynamic range. Image used courtesy of Tower Semiconductor
Aside from CMOS sensors, Tower Semiconductor has showcased optical phased arrays as another way to improve LiDAR systems. These devices are composed of hundreds of optical antennas in a compact footprint along with amplitude and phase modulators that fit neatly onto a single silicon chip. This solution delivers controllable scanning patterns, speed, and resolution for automotive cameras to assist sensors.
Tower Semiconductor has also launched an open foundry Silicon Photonics (SiPho) platform. This platform features low-loss silicon nitride waveguides, germanium photodetectors, and PIN-diode-based phase shifters. These shifters are said to enhance LiDAR systems in low-light environments at longer distances.
Integration Advances LiDAR SoCs
Beyond extending detection distances and advancing detection precision in bad weather, LiDAR developers also face the question of integration. To this end, Omni Design and LeddarTech have recently announced next-generation SoCs geared to help manufacturers design solid-state LiDAR products.
LeddarTech’s LiDAR SoC uses Omni Design’s analog-to-digital converters (ADC), plus a multi-channel analog front-end (AFE). Omni’s ADC incorporates its patented SWIFT technology, op-amp architectures for signal processing through open-loop gain, extended bandwidth, noise mitigation, and offset voltage options.
The AFEs are coupled with the ADCs on a single IP block to help prevent bottleneck from all the images and data that is collected in real-time by the LiDAR systems.
LeddarTech’s LeddarCore helps with high-data acquisition while being the central command of the LiDAR system. Image used courtesy of LeddarTech
LeddarTech utilizes Omni Design’s sensor platform to address a few limitations that current LiDAR SoC solutions face—the main one being discrete devices on PCBs. These external mixed-signal circuits, op-amps, and optical amplifiers can lead to high power consumption, data leakage, and inaccurate image processing in LiDAR systems. LeddarTech’s integrated LiDAR platform avoids external components and can be incorporated into multiple existing LiDAR architectures.
The LeddarCore 3D (LCA3) SoC is a multi-channel (64-input channels) receiving and sequencing solution that has a range of 300 meters. Manufacturers will also have LeddarTechs’ LeddarEngine, a sensor platform that supports multiple LiDAR applications and can process nearly five billion samples per second.