The basic maneuvers are prior art; the claim lives in the circumstance.

An examiner can assemble most driver assistance claims from systems already in the record: one reference that recognizes the lane, another that changes lanes, a third that warns of a collision. Under KSR, combining familiar elements by known methods is likely obvious when it yields only predictable results. What usually survives is the pairing of a particular driving circumstance — a branch in the road, the geometry of a lane marking, an obstacle beside the vehicle at low speed, a driver request the control should moderate — with the control response that circumstance calls for.

That pairing has to be explained in the specification, not merely recited in the claim. The argument against a combination of references is built from the application's own account of why the known approach fails in that circumstance and what the claimed response does about it; a specification that describes the control and never the problem leaves the claim to rest on attorney argument alone. A claim tied to one circumstance is also easy to avoid, so the protection comes from a family of applications covering the circumstances a competitor's system will actually encounter.

Vehicle control applications are commonly filed with parallel independent claims to a device, a method, and a storage medium, each reciting the same steps. The device claim is the exposed one. Written as a series of units each configured to perform a function — a recognition unit, a determination unit, a control unit — it can fall under Section 112(f) even though the word "means" never appears: since Williamson the presumption against that treatment is no longer a strong one, and a unit defined only by what it does names no structure. The claim then covers only the structure the specification discloses for each function, and its equivalents; for a function a computer performs, that structure is the algorithm, and a specification that shows a labeled box with no algorithm behind it leaves the claim indefinite.

Examiners are directed to identify these limitations and state the interpretation on the record, and the response is a decision about scope, not a formality: accept the construction, or amend to recite structure — the processor and memory that carry out the steps. The specification decides either way, which is why each functional unit needs its algorithm written out — the conditions, the thresholds, the order of the determinations — in the text and the flowcharts. The method and storage-medium claims are ordinarily not construed under Section 112(f), so without that care the device claim can end up narrower than the method claim filed beside it.