# How Chip-Scale Radar Is Shrinking Autonomous Systems

An episode of [The Hardware Podcast with Fexingo: Chips, Devices, and Electronics Engineering Conversations](https://invisiblenetwork.ai/podcast/1896799982), 1 Sep 2026, 9:36.

Audio: https://audio.fexingo.com/business/the-hardware-podcast/episode-0173.mp3

We explore how millimeter-wave radar has evolved from bulky automotive safety gear into a pocket-sized sensor for drones, robots, and smart homes. Lucas and Luna break down the engineering leap that allowed Texas Instruments and Analog Devices to shrink high-frequency signal processing onto silicon. We look at why optical sensors fail in fog and dust, but radar sees through it all. The conversation covers the specific trade-offs between resolution and power consumption, and why this shift matters for the next generation of autonomous devices operating in September twenty twenty-six. #ChipScaleRadar #MillimeterWave #AutonomousSystems #TexasInstruments #AnalogDevices #SensorFusion #EmbeddedEngineering #HardwarePodcast #TechEngineering #DroneTechnology #IoTDevices #SignalProcessing #SiliconValley #Robotics #FexingoBusiness #BusinessPodcast #ElectronicsDesign #SupplyChain Keep every episode free: buymeacoffee.com/fexingo

## Transcript

Lucas: If today's deep dive into chip-scale radar gave you a clearer picture of what’s actually inside your next drone or robot, consider supporting the show directly. A small group of listeners already funds these episodes through buy me a coffee dot com slash fexingo, keeping us ad-free and focused on the engineering details rather than clicks.
Luna: Exactly. It’s just a way to keep the lab coat on and the ads off. So, let’s get back to those tiny chips.
Lucas: Right. For years, if you wanted radar, you bought a box the size of a toaster that weighed five pounds. It was strictly for cars, bolted to bumpers, dealing with massive heat dissipation. But look at what happened over the last thirty-six months: companies like Texas Instruments and Analog Devices managed to squeeze a full millimeter-wave transceiver, including the phase-locked loops and the analog to digital converters, onto a single package smaller than a fingernail.
Luna: And that’s not just about making things fit in a drone; it’s about cost. You’re talking about dropping the bill of materials from hundreds of dollars to under fifteen.
Lucas: Precisely. And here is the concrete detail most people miss: it’s not just size. It’s the frequency band. We are now seeing reliable operation at sixty gigahertz on these chip-scale modules. That is significantly higher than the traditional seventy-seven gigahertz automotive bands, which allows for much finer resolution without requiring a huge antenna array. You can get centimeter-level precision with four antennas instead of forty.
Luna: So why didn’t we see this earlier? Wasn't sixty gigahertz just too noisy for consumer electronics?
Lucas: It was. The atmospheric absorption is higher, and the noise floor in a smartphone or laptop environment is messy. But the breakthrough wasn't just the silicon process node; it was the digital backend. These new chips handle a lot of the clutter cancellation in hardware before the data even hits your microcontroller. They filter out the static noise of the device itself.
Luna: That changes the whole architecture. You don't need a heavy-duty processor just to clean up the signal anymore.
Lucas: Correct. Which brings us to the biggest use case outside of cars: robotics. In September twenty twenty-six, we are seeing a massive shift in warehouse automation and outdoor delivery bots. Optical cameras and LiDAR are great, but they choke in fog, dust, and direct sunlight. Radar does not care about lighting conditions. It sees through dry fog where LiDAR fails completely.
Luna: I saw a demo last week where a delivery bot kept moving through a dusty construction zone while the camera-equipped rival just froze. It was basically blind.
Lucas: That is the classic failure mode for pure vision systems. Radar provides a robust fallback layer. But the real innovation here is 'through-wall' capability. Because these chip-scale radars operate at lower frequencies or have enough dynamic range, they can detect motion through drywall or thin wood. This is opening up entirely new categories for indoor positioning systems.
Luna: Wait, through walls? So your smart home could track exactly where you are standing in another room without a camera?
Lucas: Yes, and that is both a privacy goldmine and a privacy nightmare. Imagine a hospital system that tracks nurse locations for efficiency without wearing badges. Or a fall-detection system for elderly care that works through bathroom doors. The resolution isn't high enough to identify faces, but it is perfect for tracking gait and position.
Luna: But doesn't the resolution drop off sharply when you try to go through multiple obstacles? Signal loss must be brutal.
Lucas: It is. The attenuation increases with frequency. That’s why these chips are often paired with ultra-wideband radio for coarse location, and radar for fine motion detection. It’s a fusion problem. You use the cheap UWB for general area, and the millimeter-wave radar for precise velocity and micro-movements, like breathing.
Luna: So it’s not replacing other sensors, it’s becoming the glue that holds multi-sensor systems together.
Lucas: Exactly. And look at the power draw. A ten-year-old automotive radar module pulled thirty watts. These new chip-scale units pull less than half a watt. That makes them viable for battery-powered IoT devices that need to run for years on coin cells or small lithium packs.
Luna: Half a watt is huge for environmental sensors. You could put one in a forest to detect animal movement without changing batteries every month.
Lucas: That is literally one of the pilot programs happening right now in Scandinavia. They are using these chips to monitor wildlife corridors. No cameras, no IR flashes, just silent, passive radar pings that detect movement through dense foliage.
Luna: That feels like a much more ethical application than tracking people through walls. Less creepy, definitely.
Lucas: Fair point. But the tech is neutral. The same chip that tracks a deer can track a person walking behind a hedge. The engineering challenge now is software-defined radar. Instead of hard-wiring the pulse patterns, manufacturers are letting users program the waveform via firmware updates.
Luna: So you could update your drone’s radar sensitivity after you buy it? That seems risky for security.
Lucas: It is a risk, but also a feature. If a new interference pattern emerges from nearby Wi-Fi routers or Bluetooth devices, you can patch the radar’s filtering logic remotely. Static hardware can’t adapt to a changing RF environment. Dynamic firmware can.
Luna: I suppose that makes sense. But doesn't that require a constant internet connection for the device?
Lucas: Not constantly. Most updates happen over-the-air during idle times, or the initial configuration is done locally. The key is that the baseband processor on the chip is powerful enough to run these complex algorithms in real-time. It’s no longer just an analog front-end dumping raw data to a CPU.
Luna: So the intelligence is moving closer to the antenna. That reduces latency significantly.
Lucas: Massively. For a self-driving bike or a fast-moving robot, waiting for data to go up to a cloud server and back is unacceptable. Millisecond delays matter. Processing it on the chip means you react instantly to a sudden obstacle.
Luna: Which brings me back to the cost argument. If the processing is on-chip, do we still need expensive GPUs in these devices?
Lucas: For pure radar tasks, no. But for perception, you still want a GPU for visual context. However, the radar acts as a trigger. It tells the GPU, 'Hey, there’s something moving left at three meters, focus your attention there.' It saves the GPU from processing empty space.
Luna: That’s efficient computing. Filtering out the noise before it even reaches the main brain.
Lucas: Exactly. It’s about computational efficiency. We are moving away from brute-force processing toward targeted sensing. And this trend is spreading beyond mobility. Think about smart mirrors in bathrooms that can detect heart rate variability by measuring chest movements through the skin. That’s radar.
Luna: Vital signs monitoring without contact patches? That sounds incredibly useful for health tech.
Lucas: It is. And again, privacy-friendly because it doesn’t record audio or video. Just motion data. But the supply chain for these chips is tight. The fab capacity for specialized RF processes is limited compared to standard logic nodes.
Luna: So availability might be the bottleneck, not the technology itself.
Lucas: Spot on. Companies like GlobalFoundries and TSMC are expanding rf specific lines, but it takes time. If you are designing a product for late twenty twenty-six, you need to secure those allocations now. The demand from the IoT sector is outpacing the fab expansion.
Luna: A classic hardware startup trap. Great tech, no chips to build it with.
Lucas: Always. But for those who solve the supply chain puzzle, the opportunity is enormous. We are looking at a market that could hit several billion dollars in annual revenue for chip-scale radar components alone within five years.
Luna: From niche automotive parts to essential IoT infrastructure. The trajectory is clear.
Lucas: It is. And as the algorithms get better at interpreting the clutter, the hardware requirements will actually simplify. We might see even fewer antennas needed in the future, driven purely by smarter signal processing.
Luna: So the chip gets smaller, but the mind behind it gets smarter. I like that progression.
Lucas: That’s the story of modern electronics. The physics limits us, but the software keeps pushing the boundary. What do you think? Will we see these in our everyday appliances soon, or is that still too invasive?

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.