Due to the dynamic and uncertain nature of the car environment, a massive amount of local computing power and advanced algorithms to enable quick adaptation and timely decision-making is deemed essential. At the same time, reliable sensors to feed the driving system are crucial to making a fully informed decision.
The major part of the perception layer of an autonomous driving system is dedicated to simultaneous localization and mapping (SLAM), where the map of unknown environment is constantly updated with respect to the location of the car. SLAM algorithms operate based on inputs from a few different types sensors, and the majority of them are imaging sensors. This is rather obvious considering the fact that humans also rely heavily on the visual information while driving. Imaging sensors commonly considered for self-driving cars are camera, RADAR, and LiDAR
LiDAR (Light Detection and Ranging) captures high-definition real-time 3D images of the surrounding environment through active sensing with infrared lasers. It has unique advantages that can compensate the fundamental limitations in camera-based 3D imaging via vision algorithms or RADARs, which makes it an important sensing modality to guarantee robust autonomy in self-driving cars. Cameras are fundamentally passive sensors relying on ambient light, and thus the imaging is signicantly limited at night or under bad weather conditions.
Lidars (Light Detection and Ranging) are similar to radars in that they operate by sending light pulses to the targets and calculate distances by measuring the received time. Since they use light pulses that have about 100,000 times smaller wavelength than radio waves used by radar, they have much higher resolution. Distance traveled is then converted to elevation.
However, the high price tag of existing commercial LiDAR modules based on mechanical beam scanners and intensity-based detection scheme makes them unusable in the context of mass produced consumer products. Integrated coherent LiDAR with optical phased array-based solidstate beam steering, which has great potential to dramatically bring down the cost of a LiDAR module
LIDAR (Light Detection and Ranging) is considered to be the key technology for self-propelled vehicles. Mobility analysts, urban planners and AI companies bill widespread lidar as a building block for future urban societies, where autonomous vehicles, smart homes and infrastructure work together to create “smart” cities. Autonomous driving has the potential to turn the entire classical supply chain in the automotive industry upside down or to have completely new players appear on the scene. The Intel company Mobileye was recently allowed to test-drive the system in the Bavarian capital Munich. Intel bought the company in 2017 for 15.3 billion US Dollars. Amazon has just bought the startup Zoox, a company for autonomous driving, for 1.2 billion dollars.
LIDAR data is both high-resolution and high-accuracy, enabling improved battlefield visualization, mission planning and force protection. LIDAR provides a way to see urban areas in rich 3-D views that give tactical forces unprecedented awareness in urban environments.
Modulation and Detection Schemes
In intensity-modulation, direct-detection (IM-DD) LiDARs, the intensity of the source laser is modulated, and the TOF is estimated by comparing the intensity patterns of transmitted (TX) and received (RX) light in the time domain. Pulsed LiDAR, or direct TOF sensing, is the most well-known and intuitive variant where short pulses of light are transmitted and then the time-domain gap between TX and RX pulses are directly measured using an electronic timer.
In coherent LiDARs, modulation is done in the laser phase/frequency domain while the intensity is often kept constant. The reflected light is optically mixed with the local oscillator (LO) laser, which is typically realized by simply tapping-of certain portion of the transmitted laser. Finally, the TOF is inferred from the downconverted electrical signal at the optical mixer output.
Frequency modulated continuous wave (FMCW) LiDAR, a popular coherent LiDAR variant, where the frequency of tunable continuous-wave laser is linearly modulated (a triangular wave is used in this example). One can notice that the TOF causes instantaneous frequency difference between two lasers (∆f ). This difference, or beat frequency, is
linearly proportional to the TOF. Namely, by recording the beat frequency, the distance to the target is measured.
The low power transmit chirp (green) is reflected off an object. The frequency shift between the returning chirp (blue) is proportional to the distance and velocity of the object. An up and a down chirp are used to resolve for both values, distance and velocity.
FMCW LIDAR Technology
Chirped FM lidar modulates the phase of the light source (usually a single-mode laser) such that the optical frequency of the light source is modulated directly. A free-space path encodes a phase shift on the optical chirp, and the phase shift is detected by mixing the reflected chirp with a non-delayed version of the chirp. This mixing occurs at the photodiode upon detection, so no special design beyond good detector design is needed to achieve this mixing effect.
Comparison with TOF LIDAR
Most common Time of Flight (ToF) LIDAR systems operate at wavelengths of 850 and 905 nm, which are very close to the visible light spectrum. The maximum laser power is therefore limited and the range is usually less than 100 m. When driving autonomously, however, it is essential to detect objects at distances of 250 meters and more in order to brake in time or initiate an evasive maneuver. The higher the range of the LIDAR system, the more time it takes the car or driver to react to unexpected obstacles. Even fractions of a second make a big difference in terms of safety and comfort.
Frequency Modulated Continuous Wave (FMCW) LIDAR according to experts, will facilitate the breakthrough in autonomous driving and, will completely force the currently used Time of Light (TOF) LIDAR systems out of the market. FMCW RADAR is already being offered as a standard driver assistance system in the automotive industry. In contrast to radar, FMCW LiDAR does not use radio waves but light in the form of lasers, thus enabling better image resolution and improved detection of objects, such as pedestrians and cyclists.
This FMCW LiDAR technology has significant advantages over conventional methods.
- Range of 250 m+ and very robust against bad weather conditions (e.g. fog, snowfall) or direct sunlight
- Immune to sensor cross-talk and self-interference: Light impulses from other sensors cannot be confused or disturbed by own, previously sent impulses.
- Simultaneous measurement of distance and speed in each data point and thus reduced computing effort and system cost
- Low cost and scalability by using highly integrated PiCs (Photonic integrated circuits)One of the reasons why FMCW LIDAR technology is so powerful is that the sensors used can detect the smallest possible amount of light – a photon. FMCW LIDAR operates in the wavelength range of 1550 nm, thus the laser meets the required high standards of eye safety. In addition, it can provide accurate measurement results even under limited visibility conditions, for example in fog, rain or snowfall.
However, other LiDAR companies such as AEye believes that high shot-rate, agile-scanning TOF systems serve the needs of autonomous vehicle LiDAR more effectively than FMCW when cost, range, performance, and point cloud quality are important. One Argument is that FMCW cannot measure lateral velocity simultaneously, in one shot, and has no benefit whatsoever in finding lateral velocity over ToF systems.
Scantinel FMCW Sensor
By using proprietary linear chip technology, the Scantinel FMCW Sensor can instantly measure both distance and speed of any measurement point. The Scantinel system is designed to respond only to its own light pulses due to the coherent measurement process. If the returning light does not match the originally emitted light, the FMCW sensor can filter out this data point. The incoming data is also processed more quickly, since the speed from changes in object position no longer needs to be estimated, as with ToF systems, which reduces computing power and cost. By far the most important advantage of the Scantinel technology, however, is the integration of all components onto a single chip, a so-called PIC (Photonic integrated Circuits), to achieve the ambitious cost targets of LIDAR systems.
Chip-integrated waveguides for wavelengths in the range of 1550 nm are already being used by the millions in telecommunications and opto-electronic applications in data centers and are therefore available on the market at very low cost. They also do not require moving elements, such as in the so-called MEMS based LIDAR technology, which makes the Scantinel system much more robust and less susceptible to interference. Scantinel has a successful FMCW LIDAR demonstrator in customer use and is in discussions with all leading automotive and system suppliers.
NeoPhotonics Announces Tunable, High Power FMCW Laser and Semiconductor Optical Amplifier for Coherent Lidar in Autonomous Vehicle and Industrial Sensing Applications
-NeoPhotonics Corporation (NYSE: NPTN), a leading developer of silicon photonics and advanced hybrid photonic integrated circuit-based lasers, modules and subsystems for bandwidth-intensive, high speed communications networks, today announced a new, tunable high power FMCW (frequency-modulated continuous-wave) laser module and high power semiconductor optical amplifier (SOA) chips. Both components are optimized to enable long range automotive lidar and high resolution industrial sensing applications. The FMCW Laser is C-band tunable and can be directly modulated to provide >21dBm (126mW) fiber coupled power and a narrow linewidth FMCW optical signal. The SOA chip is designed for integration with Photonic Integrated Circuit (PIC) lidar engines and provides >23dBm optical output power.
These new high output power SOAs and FMCW lasers are based on NeoPhotonics photonic integration platform and improve sensitivity and range, which enables automotive lidar systems to “see” considerably farther than 200 meters, allowing for enhanced safety. Both products operate in the 1550 nm band, which is believed to be more “eye safe”, and are currently being sampled to key customers. In addition, tunable FMCW laser sources enable lidars with configurable operating wavelength thus further enhancing the immunity of coherent lidars to external light interference.
Coherent lidar, also called FMCW lidar, uses coherent technology to greatly increase range and sensitivity by measuring the phase of the reflected light instead of relying only on intensity measurements. Coherent technology was pioneered by NeoPhotonics for communications applications and implemented in PICs using NeoPhotonics Indium Phosphide and Silicon Photonics integration platforms. Coherent lidar systems require similar chip-scale manufacturing to reduce costs and enable high volume.
Coherent detection, whether for lidar or Communications applications, uses photonic integrated circuits (PICs) to extract phase and amplitude information from the optical signal. Narrow linewidth and low phase noise lasers are required for precise phase measurements and high optical power is required to compensate for optical loss in the Silicon Photonics optical chips and to provide a sufficient return signal from distant objects for efficient detection. NeoPhotonics narrow linewidth laser and SOA can be used together or separately to optimize the lidar module performance.
“We are excited to apply our high volume photonic integration coherent technology, which we have honed for over a decade, to the adjacent market of lidar and autonomous vehicles,” said Tim Jenks, Chairman and CEO of NeoPhotonics. “The benefits of coherent technology and the physics enabling it mean we can bring the same benefits to customers in these new markets that we have brought to communications customers for many years,” concluded Mr. Jenks.