# How Chip-Scale LiDAR Is Shrinking Autonomous Sensors

An episode of [The Hardware Podcast with Fexingo: Chips, Devices, and Electronics Engineering Conversations](https://invisiblenetwork.ai/podcast/1896799982), 29 Jul 2026, 7:15.

Audio: https://audio.fexingo.com/business/the-hardware-podcast/episode-0139.mp3

Traditional spinning lidar is bulky and expensive. This episode explores how chip-scale lidar—integrating laser, scanner, and receiver onto a silicon photonics chip—could bring autonomous sensors to mass-market cars, robots, and drones. We discuss the key technologies: optical phased arrays and frequency-modulated continuous wave (FMCW), recent breakthroughs from MIT, and the cost trajectory from thousands of dollars to under $100. Listeners learn why solid-state lidar could unlock autonomy in everyday devices. #Technology #HardwarePodcast #LiDAR #ChipScale #SiliconPhotonics #OpticalPhasedArray #FMCW #AutonomousVehicles #AutonomousDriving #SensorTechnology #MIT #Photonics #Robotics #DroneTech #Semiconductor #FexingoBusiness #BusinessPodcast #Engineering Keep every episode free: buymeacoffee.com/fexingo

## Transcript

Lucas: Traditional lidar has always posed a physical problem. To get the range and resolution needed for autonomous driving, you need a spinning assembly roughly the size of a coffee can. That works for a test vehicle, but it's never going to be cheap or compact enough for mass-market cars. Enter chip-scale lidar—the entire optical system integrated onto a single silicon photonics chip.
Luna: So we're talking about replacing that spinning unit with something you can barely see?
Lucas: Exactly. Some prototypes are smaller than a fingernail. The key enabler is an optical phased array—a grid of tiny antennas on the chip that steers a laser beam electronically, with no moving parts. Researchers at MIT's Photonic Systems group recently demonstrated a chip that can detect objects at 100 meters with centimeter-level accuracy. That's a huge leap from just a few years ago when chip-scale lidar could only reach a few meters.
Luna: And power consumption? A spinning lidar unit draws dozens of watts.
Lucas: Right. The MIT chip consumes less than a watt for the optical part. The overall system including electronics is still higher, but the trend is promising. There's also another approach—frequency-modulated continuous wave, or FMCW, which uses the Doppler shift of reflected light to measure velocity directly. Companies like Aeva and SiLC Technologies are commercializing that on-chip.
Luna: So two competing technologies: optical phased arrays and FMCW. Which one wins?
Lucas: It's early. Optical phased arrays give you a 2D image directly, but FMCW is more robust in sunlight because it filters out background light. Both are being integrated into standard silicon photonics processes, which means they can leverage existing chip manufacturing to drive costs down. That's the real story: lidar that can be made in the same fabs as your phone's camera chip.
Luna: Cost is the big barrier. A traditional spinning lidar unit can cost upwards of $10,000. What's the target for chip-scale?
Lucas: Several companies aim for under $100 at scale. For example, Ouster has been pushing solid-state lidar below $500 already, but the on-chip versions could go even lower. The semiconductor economics are compelling: once you've designed the chip, each additional unit costs pennies in silicon. That could make autonomy affordable for mass-market electric vehicles in the next five years.
Luna: But range and resolution have to be good enough for highway speeds. Is 100 meters enough?
Lucas: At highway speed, you need to see about 200 meters ahead to stop safely. So 100 meters is on the edge. But remember, chip-scale lidar will likely be one sensor in a suite—alongside radar and cameras. And these chips are improving quickly. A startup called SiLC Technologies claims their FMCW chip can reach 200 meters using a coherent detection technique that also measures velocity per pixel.
Luna: That velocity measurement is key for predicting movement. How does FMCW achieve that?
Lucas: FMCW sends out a laser whose frequency ramps up over time. When the reflected light returns delayed by the round trip, it interferes with the current signal, creating a beat frequency that encodes both range and velocity. It's like radar but with light. This makes it very effective for tracking fast-moving objects like other cars or cyclists.
Luna: And all of that is now on a chip. So where are the first commercial deployments likely?
Lucas: Probably in indoor robots and drones first, because the regulatory hurdles are lower. A warehouse robot doesn't need to worry about weather or sun glare. But automotive validation takes years. I'd expect to see chip-scale lidar in robo-taxi fleets around 2028 or 2029, and maybe in consumer cars by 2030.
Luna: That timeline depends heavily on cost reduction. And that's where semiconductor volume comes in.
Lucas: Exactly. The more industries adopt these chips, the faster costs drop. Think about drones, robotics, and even AR/VR depth sensing. Each market adds volume, which brings down the price for automotive. It's a virtuous cycle.
Luna: It reminds me of how we think about this podcast—if the value we deliver is worth a coffee a week, that's the kind of support that keeps us ad-free and focused on topics like these.
Lucas: Yeah, honestly, if today's hardware conversation gave you something usable, that's the link—buy me a coffee dot com slash fexingo. It's a small gesture that makes a huge difference for us.
Luna: We appreciate everyone who supports the show that way. So, back to the chips—where do you think the first volume deployments will happen? Warehouse robots?
Lucas: Likely yes. Companies like Amazon Robotics already use lidar for navigation, but they currently use spinning units. Switching to chip-scale would allow smaller, lighter, cheaper sensors that consume less power. That's a no-brainer for fleet operators. Drones are another big opportunity—especially in agriculture and delivery.
Luna: What about the AR/VR market? Depth sensing is crucial for mixed reality headsets.
Lucas: That's actually a fascinating application. Chip-scale lidar could provide high-resolution depth maps at low power, which is ideal for headsets like Apple's Vision Pro or Meta's next-generation Quest. Current depth sensors use time of flight infrared cameras, but lidar gives much better accuracy in daylight. So we might see chip-scale lidar in consumer electronics even before cars.
Luna: And all of that relies on silicon photonics. How mature is that manufacturing process?
Lucas: Silicon photonics has been in development for over a decade for high-speed data center transceivers. Companies like Intel and Cisco have been making photonic chips for years. Now we're repurposing those same production lines for lidar. That's a massive advantage—they're not building new fabs, just adapting existing ones. That keeps R&D costs down.
Luna: So we're on the verge of lidar becoming commodity hardware, much like camera sensors have become.
Lucas: Precisely. The same way CMOS image sensors enabled cheap cameras in every phone, chip-scale lidar could enable autonomy in everything from vacuum cleaners to tractors. That's a hardware revolution that's quietly underway.
Luna: And we'll keep tracking the breakthroughs. Thanks for diving into this with us.
Lucas: Absolutely. It's one of those areas where the physics and engineering are coming together to solve real-world problems. And it's happening faster than many people realize.

Shows are listed from Apple Podcasts' charts and directory, and Podcast Index where it is used. Every show and episode belongs to its publisher and plays from the publisher's own feed; this site hosts none of them.