Start with the failure mode the claim fences against: a LiDAR that drifts out of calibration mid-trip. Torc Robotics' grant US12650501B2, "Systems and methods for online Doppler LiDAR calibration for road vehicles" (issued June 9, 2026), claims calibrating the sensor online — while the vehicle is driving — rather than only in a garage. The anticipated problem is real: a long-haul truck's sensor suite shifts with temperature, vibration, and time, and a perception stack reading a miscalibrated LiDAR is reading a lie.
The CPC nails the modality. G01S 7/4972 and G01S 17/58 are about LiDAR systems that measure relative velocity — Doppler, or frequency-modulated continuous-wave, LiDAR — and B60W 60/001 ties it to autonomous vehicle operation, with B60W 2554/20 and 2554/804 covering the surrounding-object context. This isn't generic time-of-flight LiDAR; it's the velocity-sensing kind. That distinction is the whole point of the claim.
Here's the camp-lines read. Doppler LiDAR gives you per-point radial velocity in a single shot — you know not just where a thing is but how fast it's moving toward or away from you, instantly. A camera-only stack cannot get that directly; it infers velocity across frames, which is exactly the edge case that bites in fast-closing highway scenarios. By patenting online calibration of Doppler LiDAR, Torc is doubling down on a sensing capability that defines the mapped-LiDAR camp and that vision-only architectures structurally forgo.
For claim scope, the calibration angle is shrewd. The hard, ongoing problem in any LiDAR deployment isn't the first calibration — it's keeping calibration true over a vehicle's life. A grant on doing it online, in motion, fences a maintenance-and-reliability capability that a fleet operator actually needs, not just a lab demo. That's a more durable kind of claim than one on a perception trick, because the problem doesn't go away as the models improve.
Caveats in the house style. A calibration claim is bounded by its recited method — the specific signals and procedure used to estimate and correct the calibration — so a competitor calibrating by a different observable may sit outside. Velocity-sensing LiDAR remains costlier than camera, and a patent on calibrating it says nothing about whether the economics close. And, as always, classification points; claim 1 governs.
The takeaway for the autonomy-software beat: calibration patents are easy to overlook and strategically heavy. Torc's grant is a sensor-camp commitment dressed as a maintenance feature — verifiable on PatentBear, classified squarely in Doppler-LiDAR sensing, and pointed at the velocity capability that draws the line between the two autonomy camps.