The first step in designing a 3D LiDAR (light detection and ranging) sensor is to pick a laser source for generating photons to illuminate the scene. Each LiDAR has a receiver which collects and processes the light reflected back from the targets to provide real-time 3D depth information without ambiguity.
The laser is a critical piece of the puzzle for any LiDAR sensor’s architecture and most manufacturers today are using legacy components. At Sense, we’ve created a new digital light source, the Sense Illuminator, which is fundamentally better for two reasons:
1) It’s less expensive to produce
2) It illuminates the entire scene uniformly at once without scanning either mechanically or electronically.
In existing LiDAR solutions, companies source off-the-shelf laser components which shine IR light at a tiny spot, much like a laser pointer you might use to play with a cat. Their products use multiple lasers to illuminate a ‘column’ or ‘row’ of spots and then mechanically scan the scene to cover the full desired field-of-view (FOV). Since they don’t capture full-frame data, these companies have to keep track of the timestamp for each pixel that is returned per frame. Then, they use that information to correct for motion blur if a target, such as a high-speed vehicle, has moved within that frame.
But why do these companies scan their laser to begin with?
The answer is in the laser source; their innate need to scan only exists because their laser can only illuminate that tiny spot rather than the desired FOV so they have to move the laser spot across the FOV for full coverage. This causes inherent trade-offs between range, resolution, frame rate, and FOV.
It also increases system complexity, adds expensive optical components, and introduces tedious calibration and alignment procedures during assembly. As a result, it burdens customers with high-cost sensors with many potential points of failure that struggle with reliability and durability in real world use cases.
Most end customers that actually use these mechanical sensors in automotive or industrial markets are well-versed with these failures—especially in spinning LiDARs where the internal bearings can fail frequently and the sensors have to be refurbished or replaced. In addition, mechanical sensors—whether spinning, MEMS-based, or utilizing galvanometers—have trouble maintaining depth accuracy during high amplitude vibration pulses, such as those experienced on the road throughout the life of a vehicle.
Visual illustration of a legacy scanning LiDAR sensor (spinning architecture)
Recognizing all such limitations of a scanning LiDAR, Sense took a fundamentally different approach. We knew that the optimal long-term architecture to enable mass-market LiDAR would be one that did not need to scan at all. So our radically simplified architecture uses an application-specific Sense Illuminator—a “LiDAR floodlight” of more than 10,000 lasers—to fire photons at the entire FOV at once. This eliminates the need to mechanically or electronically scan the laser and allows us to create a 100% solid-state, camera-like architecture with no moving parts.
Visual Illustration of Sense Illuminator's 'floodlight effect' (global shutter flash architecture)
Our differences run deep. While many LiDAR companies rely on inefficient edge emitting laser diodes or high-cost fiber lasers, we chose to design our arrays with custom vertical cavity surface emitting lasers (VCSELs) and shrunk the size of each VCSEL die to less than the width of a hair strand. In other words, we can pack 250 VCSELs in an area equivalent to the head of a pin!
To build these arrays, we first grow millions of VCSELs on a single wafer and then use a patented technique called micro-transfer printing (MTP) to “print” thousands of these VCSELs in a single step to a heat-conducting flexible substrate. Our MTP technique distributes these lasers in a thermally optimized manner allowing us to output kilowatts of peak optical power while maintaining class 1 eye safety and consuming very little average electrical power.
Animation of our micro-transfer-printing process to assemble VCSEL arrays
The unique attribute of flexibility also allows for a simple way to tailor the horizontal FOV by controlling the amount of bend applied to the Sense Illuminator. Each die on our array operates at 940nm, a unique wavelength with very little solar flux output, to eliminate interference from the sun.
VCSELS became popular in data communication applications because they’re lighter, easier to fabricate, more reliable, have less noise, and maintain operation at higher temperatures than traditional pulsed lasers. For these very reasons, we’re now also seeing VCSELs increasingly applied in consumer electronics. Earlier this year, Apple launched a VCSEL-powered LiDAR embedded inside an iPad Pro and more recently in October 2020, added LiDAR sensors in the iPhone 12 Pro for creative applications such as photography and augmented reality.
As demand for LiDAR sensors increases, VCSELs will drive system costs down significantly with production scale. And as we’re already witnessing, VCSELs are beginning to follow Moore’s law of performance improvements. At Sense, we’ll take advantage of these improvements and continue to shrink our VCSEL dies to make denser and denser “LiDAR floodlights” that are a core building block of our high-performance, low-cost, and reliable LiDAR sensors.
Talk to our experts for more information about our products and technology.