The first low-cost, compact, semiconductor-based LiDAR sensors
With their innovative LiDAR OS–1 series, Ouster achieves incredible performance.
In this article, we look at the disrupting technology which sets Ouster to lead the market with a portfolio of low-cost, compact, semiconductor-based lidar sensors in both scanning and solid-state configurations.
When studying the OS-1-16 specifications, its performance, its characteristics and evaluating all of this against its pricing, one thing that strikes is: “how can, a sophisticated lidar as such, achieve such good performance, be so light and compact, collect ambient imagery in addition to lidar images, and yet, cost much less than other comparable products of this type? “
The simple answer is that the technology behind this lidar is not the same as the one used by other manufacturers. In this article, we will examine a little more in details, what the technology behind is, and how it enables this sensor to potentially be working on all cars and outdoor robots.
To do so, we start by explaining what is behind the unusual term “Multibeam Flash” technology. It primarily relies on an uncommon usage of 850nm laser wavelength. Most lidar systems operate towards slightly higher spectrums, this is so, to overcome the effects of ambient sunlight, which is more evident at lower wavelengths. This a fundamental requirement needed to see obstacles better.
At 850nm, the sunlight is 3-10 times stronger than commonly used wavelengths. Ouster devised a unique patented technology which revolves around ambient-light rejection. To put it in simple words, one can think of it as Ouster, patented a special type of sunglasses for their sensors, which allow seeing well in a range of light which, otherwise, gives an enormous deal of problems. At the same time, by being equipped with this technology, it helps their lidar operating in a wavelength range that reveals more than other sensors would see by operating in spectrums where sunlight is less intense. This gives a lot of advantages:
- Less laser energy attenuation due to air humidity
When operating at 850nm, the laser energy is absorbed less by the water vapor present in the air, this allows a more consistent and steady performance even when changes in air humidity occur.
- Longer range and higher resolution
When operating at lower wavelengths, the light reflected is captured better, the CMOS can see circa two times more. More light means that the range of usage is longer (120 m @ 80% reflectivity for the OS-1-16) than other lidars and that the resolution is higher (1,2cm for OS-1-16).
- High-quality ambient imagery
The firmware for OS-1 outputs fixed resolution depth images, signal images, and ambient images in real time, all without a camera. The data layers are perfectly spatially correlated, with zero temporal mismatch or shutter effects, and have 16 bits per pixel and linear photo response. The image capturing process is made without the need for a camera because, at this wavelength range, more photons from the sun are available to capture the imagery or the environment around. Furthermore, boosting this information is easy because, at this spectrum, there is a high signal-to-noise ratio, making it ideal for capturing images at dawn, dusk or when cloudy.
- Focus on efficiency
The flood illumination in a conventional flash lidar, while simpler to develop, wastes laser power on locations the detectors are not looking. By sending out precision beams only where detectors are looking, Ouster achieves a major efficiency improvement over a conventional flash lidar.
Video made with OS1-64 Lidar
The manufacturing is based on an all-semiconductor approach that leverages a unique laser-detector combination: vertical cavity surface emitting lasers (VCSELs) and single photon avalanche diodes (SPADs). VCSELs and SPADs are cutting edge laser and detector technologies that are broadly deployed in markets outside of lidar. Until now there have been insurmountable challenges to using them in high-resolution lidar systems.
Multi-beam flash lidar architecture plays to the strengths of VCSELs and SPADs. In this case, “flash” refers to the idea that every pixel in the sensor is illuminated by the laser and actively collecting light simultaneously like a camera with a flash, and “multi-beam” refers to the fact that the scene is illuminated with precision beams of light instead of a flood.
Thanks to this technology, the Ouster OS-1 can output structured lidar data where horizontal and vertical angular spacing are always kept constant – just like a camera. This allows the lidar to output a fixed-size 2048 by 64-pixel depth map as well as intensity and ambient light images on every frame. This makes it the optimal visual sensor for AI systems that use special types of machine learning methods suited for analyzing visual imagery (CNN = Convolutional Neural Networks). Structured Lidar data also eases image storage and labeling workflows.
Summary
Ouster shows an impressive portfolio of low-cost, compact, semiconductor-based LiDAR sensors. This is because the technology behind this lidar is not the same as the one used by other manufacturers. Ouster devised a unique patented technology which revolves around ambient-light rejection which unleashes the potential benefit behind an uncommon usage of 850nm laser wavelength. This provides a more consistent and steady performance even when changes in air humidity occur, longer range and higher resolution. Ambient images in real time can be provided all without a camera. By sending out precision beams only where detectors are looking, Ouster achieves a major efficiency improvement over a conventional flash lidar. The Ouster OS-1 can output structured lidar data, which makes it the optimal visual sensor for AI systems that use special types of machine learning methods suited for analyzing visual imagery.
