When an automated-driving company patents how its car perceives the world, the classification usually points to one place. The Cooperative Patent Classification (CPC) symbol G01S is, in the USPTO's published class title, the heading for "radio direction-finding; radio navigation; determining distance or velocity by use of radio waves; locating or presence-detecting by use of the reflection or reradiation of radio waves; analogous arrangements using other waves." That final clause — "analogous arrangements using other waves" — is what pulls optical LiDAR (light detection and ranging) into the same class as radio-wave radar. Both measure range and velocity by emitting a wave and reading its reflection; CPC treats them as analogous.

For automotive work, two subgroups appear constantly. G01S 13/931 reads on radar systems specially adapted for anti-collision purposes in road vehicles; G01S 17/931 reads on the LiDAR equivalent. When a record carries one of these, the classification itself tells you the sensing modality and that the application is road vehicles — before you have read a single claim. Records that carry both are typically sensor-fusion inventions that combine radar and LiDAR returns.

"RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES"— USPTO, source

How a real sensor-fusion grant classifies

A granted record makes the pattern concrete. US 12,117,519 B2, "Object detection using RADAR and LiDAR fusion," assigned to Motional AD LLC and issued October 15, 2024, is classified in G01S 13/865 (radar combined with other sensors), G01S 7/412 and G01S 7/417 (radar signal processing), and G01S 13/42, among others. Its independent claim 1 walks through obtaining a LiDAR point cloud and a radar point cloud, transforming both to a common vehicle-centric reference frame, and then combining them. The claim states the fusion step explicitly:

The patent claims "generating the third set of clusters includes combining features at a feature level to create fused LiDAR/RADAR-generated features using features obtained from each of a LiDAR branch and a RADAR branch in a network." That is the enabling step a reader should locate: the claim does not merely describe two sensors, it recites combining their features inside a network to produce a fused detection. Whether a perception patent is genuinely enabling or merely aspirational often comes down to whether the independent claim recites a concrete combination step like this one, or only gestures at "fusion" in the abstract.

Why perception patents concentrate in G01S

Automated-driving perception is, at bottom, a ranging problem: estimate where objects are and how fast they are moving. Because G01S is defined around determining distance and velocity from reflected waves, it is the natural home for the radar and LiDAR inventions at the front of the autonomy stack. Adjacent classes pick up neighboring problems — G06V and G06T for the image-processing and computer-vision steps, B60W for the control logic that acts on a detection, and G05D for the autonomous-operation layer. A single perception patent frequently carries symbols from several of these at once, and the combination is itself a roadmap tell about how far up the stack the invention reaches.

For an analyst, G01S 13/931 and G01S 17/931 are the precision filters: query the patent record by these symbols plus an assignee to surface that company's road-vehicle radar and LiDAR portfolio.

Why radar and LiDAR share a class

It can seem odd that a radio-wave technology and an optical-laser technology live under the same heading, but the class title's logic is precise. G01S is organized around the function — determining distance or velocity by reflected waves — not the medium. Radar emits radio waves and times their reflection; LiDAR emits light pulses and times theirs; both solve the same ranging problem with the same mathematical core. The "analogous arrangements using other waves" clause is the bridge that lets CPC group them, which is exactly why sensor-fusion inventions that combine the two modalities fit naturally into the class rather than being split across unrelated headings. This also explains the subgroup parallelism: G01S 13/xx covers radar systems, G01S 17/xx covers LiDAR systems, and the trailing /931 marks the road-vehicle anti-collision application in both branches.

The fusion case sharpens the point. A patent that fuses radar and LiDAR will typically carry a G01S 13/865 symbol — radar combined with sensors of a different type — precisely because the invention's contribution is the combination across modalities. When an analyst sees G01S 13/865 on a record, the classification is signaling a multi-sensor architecture before any claim is read. Pairing that with the assignee tells you which companies are patenting fused perception rather than single-sensor detection.

Cautions when reading G01S as a signal

The usual classification caveats apply with force in perception, where filing volume is high. First, a G01S symbol describes subject matter, not legal scope: a published G01S application is a disclosure under examination, and only the issued claims of a grant define what it can exclude — a distinction that matters because aspirational perception applications are common. Second, perception patents are heavily co-classified; a single LiDAR record may carry G01S, G06V, G06T, B60W, and G05D symbols at once, so the lead symbol alone undersells how far up the stack the invention reaches. Third, examiners assign and revise CPC symbols during prosecution, so class counts measure activity in a sensing area rather than a fixed competitive position. Read with those cautions, G01S 13/931 and G01S 17/931 remain the sharpest entry points into the automotive perception record. The class title is published and maintained by the USPTO as the authoritative definition of the symbol.