One assignee leads the ranked list with 198 of the 789 records in scope, well ahead of the fifth-ranked filer at 20 and the tenth-ranked filer at 13. The top five assignees combined account for 387 records, or 49.0% of all records in scope, and the top ten extend that to 58.6%. Below the leading names the ranking thins out quickly into a long tail of companies with only a handful of filings each, so due diligence should focus heavily on the top names rather than treating the field as evenly spread.

Filing activity rose sharply from 7 records in 2017 to a peak of 147 in 2023, but the most recent complete-year comparison shows a decline from 122 filings in 2021 to 104 in 2024, a 15% pullback. That is best read as a cooling from a hot period rather than a field going cold, especially since publication lags filing by roughly 18 months and the 2025-2026 figures in any chart will keep rising as more records publish. Anyone using this trend to time a filing decision should treat the last two years as incomplete rather than as evidence of a slowdown.

Manipulator and robot hardware claims under IPC class B25J appear in 40.4% of the 789 records, making it the single largest technology bucket. Business, commerce and administrative data-processing claims under G06Q are close behind at 37.5%, and AI-based computing claims under G06N cover 26.9% of records. Because a record can carry more than one IPC class, these percentages add up to well over 100%, but the pattern still shows that cobot patents increasingly bundle physical hardware claims with a software or data layer on top.

Image and video recognition claims under G06V cover only 8.2% of the 789 records, the thinnest of the major IPC classes tracked in this dataset, well behind control-system classes G05B at 22.6% and G05D at 13.1%. That relative thinness suggests vision-based perception work — gesture recognition, human-proximity detection, and multi-sensor fusion for perception rather than just control — has more open claim space than the heavily filed manipulator and path-planning areas. Teams drafting in this space should still run a full clearance search, since a lower IPC share does not mean the space is empty, only less dense.

US20240308070A1, published 2024-09-19, claims a method where a user manually demonstrates a welding path along one part in a sequence of identical parts, records the start and end positions across the sequence, and the collaborative robot system calculates the welding path for every remaining part from that single demonstration plus the part count. It is a fairly specific application of motion-trajectory and path-planning technology to repetitive welding tasks rather than a broad claim over cobot path planning generally. Anyone building a similar single-demonstration, multi-part trajectory-transfer feature for welding or a comparable repetitive task should review this filing's claim scope closely before assuming their approach is clear.