The dataset shows families peaking at 808 in 2021 and easing to 701 by 2022, with further declines since. This pattern typically appears once the foundational architecture claims in a field have been staked out by early movers, leaving later entrants to file narrower refinements rather than broad new systems. It does not mean R&D has stopped; it means the patenting activity tied to core system claims has slowed, and remaining filings tend to be more incremental. Readers should also account for the roughly 18-month lag between filing and publication, which understates the true count for 2025 and 2026.

Recent-year momentum data shows a small cluster of orthopedic and surgical robotics groups, including Mako Surgical, Globus Medical, Stryker, Mazor Robotics and the Smith & Nephew group of entities, as the most consistently active filers. All of them, however, show sharp year-over-year declines in the latest tracked year, ranging from roughly -79% to -98%. Co-assignee data also shows Smith & Nephew's corporate, orthopedics and Asia-Pacific entities filing jointly, which reflects internal corporate structure rather than three separate competitors.

In this corpus, image registration and localization claims fall mostly under G06T (image data processing), which covers 990 of the 5,980 records, while SLAM-adjacent manipulator and sensor-fusion work spreads across B25J and G06N with smaller counts. Image registration typically refers to aligning preoperative imaging with the intraoperative scene, while SLAM implies continuous mapping and localization of the robot or instrument within that scene as the procedure progresses. The relatively small G06N (AI models) count of 212 suggests that fully learned SLAM approaches are still a minority of filed claims compared with more conventional registration techniques.

Yes, based on the IPC composition: A61B surgical claims outnumber G06T image-processing claims by more than five to one, and G06N AI-model claims are smaller still at 212 records. That imbalance suggests claim space around marker-less soft-tissue registration, multi-modal intraoperative image fusion and AI-driven feedback-loop control is comparatively less crowded than the core surgical-device layer. A first claim in these areas would need to specify the sensing modality, the fusion or registration algorithm, and the specific intraoperative decision it drives, since generic system-level claims are likely to run into the dense A61B prior art.

US12690933B2 is the newest representative filing in this dataset, published 2026-07-28, and it claims a robotic surgical system with a feedback loop that routes sensor data through an AI-backed control system to dynamically adjust operation mid-procedure. Because it sits at the newest edge of the corpus, it signals where claim language is heading rather than an already-litigated or widely cited position. Anyone building an AI-adjusted feedback control loop for a surgical robot should review its specific claim scope closely, since a narrowly drafted independent claim can still block a broader category of implementations even without a long citation history.