The sensor modality is in the claim, and GM's grant US11449067B1, "Conflict resolver for a lidar data segmentation system of an autonomous vehicle" (issued September 20, 2022), names it plainly: lidar. The CPC anchors it — G01S 7/4802 (lidar signal processing), G01S 17/931 (automotive lidar) — fused with G06V 20/56 road-scene recognition and G05D 1/0088 control. The independent claim's value is the arbitration step: when multiple segmentation processes disagree about what a lidar point cluster is, the resolver decides.
That is a genuinely enabling claim, because conflict resolution is a real, specific problem in a production perception stack. Mature systems run several segmentation methods — geometric, learned, temporal — and they do not always agree. A naive pipeline either trusts one blindly or stalls; a conflict resolver is the engineering answer, and claiming the resolution mechanism is claiming something the system actually has to do. This is not aspiration; it is a named sub-component.
On scope, the independent claim establishes the resolver in a lidar-segmentation context; the dependents that specify the arbitration logic — confidence weighting, temporal consistency, geometric priors — are the moat. A dependent that ties the decision to a defined confidence signal is more defensible than the broad notion of resolving conflicts. The B1 kind code means this issued without a prior publication, so the scope has been examined and stands.
Strategically, GM's Cruise-era and in-house AV work generated a steady stream of perception-component IP, and a conflict resolver is exactly the kind of unglamorous-but-essential piece a serious builder patents. Dated September 2022, it reflects a stack mature enough to have a multi-method segmentation problem worth resolving. The enabling-versus-aspirational verdict is clear: enabling, with the real protection in the arbitration-logic dependents — read those, not the abstract.