"Actor" is a loaded word in autonomy — it means something with agency, something that might do something. UATC's grant US11597406B2 ("Systems and methods for detecting actors with respect to an autonomous vehicle," issued March 7, 2023; inventors include Carlos Vallespi-Gonzalez) fences the detection of actors specifically: not all objects, but the dynamic agents whose future behavior the ego vehicle must anticipate. The framing separates a parked car from a moving one, a wall from a pedestrian.

The mechanism leans on velocity-aware LiDAR — G01S 17/58 (Doppler/velocity) and 17/66 (tracking) — plus learned models (G06N 20/00) to identify and characterize moving agents, with B60W 60/0027 tying the result to control that accounts for object trajectories. By fencing actor detection rather than generic object detection, the claim stakes out the perception sub-task that feeds prediction and planning, which is where the autonomy hard problem actually lives.

For the control-and-perception beat, the actor abstraction matters because the whole downstream stack — prediction, interaction modeling, planning — is organized around agents, not pixels. A claim at the actor-detection layer sits upstream of all of that. Fencing it is fencing the input to the part of the system that decides how to behave around other road users.

From a portfolio angle, this is another live family: it was continued as US12214808B2 (same title, 2025), so the enforceable scope is defined by the newer child. The continuation, filed as Aurora absorbed UATC's autonomy assets, is the tell that actor-detection IP stayed strategically active across the corporate transition — watch the assignment and the continuation together.

Caveats. Dynamic-object detection is heavily prior-arted; the grant turns on the specific actor-detection-and-characterization method in claim 1, not on the actor concept. With a continuation, read the latest claim for current scope. The independent claim's actor-detection step is the asset.

For the file: an actor-centric perception grant with a 2025 continuation, upstream of the prediction-and-planning stack. Pull US11597406B2 and US12214808B2 on PatentBear, read the newer claim 1, and track the family across the UATC-to-Aurora transition.