LiDAR nano-antenna array
OryxVision, with its nano-antenna array LiDAR, has identified a key challenge hindering the commercialization of autonomous vehicles: balancing detection range, accuracy, and cost. OryxVision's depth sensor can detect objects as small as 150 meters away. It operates in complete darkness, is unaffected by direct sunlight, and remains stable even in extreme weather conditions such as rain and fog. The OryxVision sensor can be seamlessly integrated around the vehicle; a single sensor array provides high-performance 360° depth detection coverage at a low cost.
OryxVision uses a long-wavelength terahertz infrared laser because this type of infrared light is invisible to the human eye and has higher power. This infrared light is also difficult for water to absorb and is not affected by clouds, fog, or strong direct sunlight. When the emitted laser light reflects back to the sensor, optics guide the reflected incident light onto a large number of extremely small rectified nanoantennas. The incident light produces an AC response in the rectified antennas; in other words, it is converted into a DC signal. The system's sensitivity is hundreds of times greater than that of traditional LiDAR. Because the antennas treat the incident light as a wave, they can also detect the Doppler effect (frequency changes caused by the relative motion of the reflected light), and thus determine the speed of other objects in or near the road.
OryxVision uses nanoantennas that are only 25 square micrometers in size, fabricated directly on integrated circuits using thin-film chip manufacturing processes. Each nanoantenna is only 5x5 square micrometers in size, and is formed into an array on a silicon wafer using thin-film chip manufacturing processes, making it quite inexpensive. This also makes it very simple to feed the signals to a machine learning system, which can classify objects in a scene, thus making scene perception more intelligent.
Floodlight (Flash) LiDAR
Floodlight LiDAR is one of the more mainstream technologies in all-solid-state LiDAR. FlashLiDAR is a non-scanning radar that emits area light and focuses on outputting 2D or 3D images. While it offers good stability and cost, its main problem lies in its relatively short detection range. For floodlight imaging LiDAR, each emitted beam is scattered across the entire field of view, meaning only a small portion of the laser light is projected onto specific points. Furthermore, each pixel in the photodetector array must be extremely small, limiting the amount of reflected light it can capture.
There are two main types of laser sources for floodlight area array LiDAR: pulsed and continuous wave, corresponding to pulse-to-flight (pToF) LiDAR and continuous-wave time-to-flight (cwToF) LiDAR, respectively. pToF LiDAR uses pulses and can achieve long-range detection (e.g., over 100 meters); while cwToF LiDAR uses continuous waves and is mainly used for short-range detection (e.g., tens of meters). Floodlight area array LiDAR is a non-scanning LiDAR, capturing the entire scene through pulses or continuous waves, rather than scanning point-by-point with a laser beam. Due to the weaknesses of detecting electron return pulses and wide bandwidth, floodlight area array LiDAR is susceptible to noise, and threshold triggering can cause measurement errors Δt. The figure below compares the parameters of cwToF LiDAR, pToF LiDAR, radar, and ultrasonic sensors.
As the theory of lidar states, under power constraints, achieving long detection distances requires a large pulse width, while achieving high detection accuracy requires a large bandwidth. The product of a simple pulse width and bandwidth is close to 1; therefore, pulse width and bandwidth are interrelated and cannot be increased simultaneously. pToF LiDAR, however, uses a gain-modulated pulsed laser source to solve the problem of limited instantaneous power in continuous-wave lasers emitting sinusoidal waves to image a target at a distance, which severely affects imaging quality and measurement range.
Multiple dense laser beams diffuse directly in all directions, illuminating the entire scene in a single flash. It functions more like a camera. The system then uses a miniature sensor array to collect the laser beams reflected back from different directions.
One major advantage of Flash LiDAR is its ability to quickly record the entire scene, avoiding the various problems caused by target or lidar movement during scanning. However, this approach also has its drawbacks. The main disadvantage of Flash area-array LiDAR is its photon budget: once the distance exceeds tens of meters, the number of returning photons is too small to allow for reliable detection. This can be improved by using grid-like structured light for illumination instead of light-covering the scene, sacrificing tangential resolution. Vertical-cavity surface-emitting lasers (VCSELs) allow them to emit thousands of beams simultaneously in different directions.
