A granted patent is enforceable coverage, so a week's grant list reads as a map of what an assignee has just locked in. In patents issued the week of April 14, 2026, Hyundai — counting Hyundai Motor (HYMTF) together with its Hyundai Mobis components affiliate — holds a sizable cluster of autonomous-driving grants, 38 in the records reviewed across the group. The notable thing is not the count but where the claims sit. They concentrate less on the core act of driving and more on the awkward, safety-critical edges: the handover back to a human, the moment a sensor reading is in doubt, and the braking path of last resort.

The handover-and-doubt bucket

The clearest example of the handover theme is US12600374B2, an apparatus for controlling autonomous driving that detects an event requiring transfer of control to the driver, issues a first and then a second escalating notification when the driver does not respond, and reduces the vehicle's speed at a defined rate if no response comes. That is a minimal-risk-condition claim — coverage over the choreography of giving up control safely when the human will not take it. US12600366B2 addresses the doubt case: when a detected vehicle meets a specified condition, the system sends a cross-check request to other vehicles and uses their existence or absence responses to judge whether its own sensor reading is real and whether its sensors are working. Its abstract states the mechanism directly.

Identify an operating state of the sensor device based on a host vehicle existence response or a host vehicle absence response among responses received from the other vehicle in response to the cross-check request.— Apparatus for controlling autonomous driving and method thereof, US12600366B2

Two more grants round out the perception-and-decision side. US12602913B2 covers identifying object priority for autonomous driving control using LiDAR contour points, sorting detected objects into groups and ranking which to display or act on first. US12600380B2 is narrower and concrete: an autonomous-driving control apparatus that recognizes a road security mirror ahead, stops the vehicle when the mirror meets a specified condition, and resumes once the image change rate in the mirror falls below a threshold — coverage over a specific blind-intersection maneuver. Both carry the B60W 60/0015 and G06V 20/588 classes that mark autonomous-operation and lane-scene recognition work.

The fallback-braking bucket

The second concentration is in the components affiliate and is about the brake of last resort. US12600334B2, an electronic brake device, explicitly claims providing redundant braking force in situations where a driver is not driving or is paying less attention — such as autonomous driving or smart cruise control — and the main braking device malfunctions. US12600349B2, a forward-collision-avoidance assist method, selects a candidate front object, determines a target object from the vehicle's predicted driving trajectory, and controls the brake based on collision risk. And US12600360B2 covers steering-control curvature, comparing a camera-sensed lane curvature against the map curvature from the navigation system before applying one or the other — a claim about reconciling what the camera sees with what the map says. These are the actuation and control claims that sit beneath the decision layer, and they issued to Mobis in the same week the decision-layer claims issued to Hyundai Motor.

Two grants in the supporting set are worth noting because they show how wide the same week's filing reached. US12601840B2 covers a LiDAR object-tracking method that clusters a point cloud in two passes — a first clustering, then a second clustering applied in response to how long the first one took. That is a claim about keeping perception within a time budget, the unglamorous real-time-performance problem that determines whether a tracker is usable in a moving vehicle at all. It pairs naturally with the object-prioritization grant: one decides what to track, the other decides how fast it can afford to. Both are Hyundai Motor claims rather than Mobis, which underscores that the group's coverage this week ran from the components affiliate's actuation layer up through the parent's perception and decision layers in a single issue cycle.

The two buckets connect into a single map: the handover, cross-check, and prioritization grants govern when and how the system decides it has a problem, and the redundant-brake, collision-assist, and steering-reconciliation grants govern what physically happens next. A group issuing across both layers in one week is building coverage around the failure-and-fallback envelope of autonomous driving — the part that determines behavior when something is wrong — rather than around the nominal driving model alone.

It is also worth noting what these grants are not. None of the cluster claims a faster or more accurate way to perceive the road in nominal conditions — the core driving competence. Every one of them activates only when something deviates: a driver who will not retake control, a sensor reading that needs corroboration, a blind intersection, a main brake that fails, a camera lane line that disagrees with the map. In CPC terms the cluster leans on B60W 50/14 (driver notification), B60W 50/0205 (fault response), and B60W 60/0015 (autonomous-operation handover) rather than the pure perception classes. A group concentrating a single week's issued claims on those deviation-and-recovery classes is building enforceable coverage around the exception path of autonomous driving. For a reader trying to size what Hyundai locked in, that is the concrete takeaway: the grants describe the parts of the system that exist for when things do not go as planned, and they issued across the Motor parent and the Mobis supplier in the same cycle.

Two caveats keep the read grounded. Grants reflect prosecution begun years earlier, so this week's issue list is coverage Hyundai sought in the past, not a freshly announced direction. And a granted claim defines what can be asserted, not a shipped feature; these records are claims, and several explicitly describe apparatus and methods rather than products in market. What the issued patents establish concretely is the footprint: in one week, Hyundai Motor and Mobis together added enforceable claims spanning control handover, V2V sensor cross-checking, object prioritization, redundant braking, and steering reconciliation, with the relevant Hyundai entity on the assignee line of each — a coverage map weighted toward the moments an autonomous system must hand back or fall back.