The most-cited record in this dataset is a 1990s anti-slip control patent for legged robots (US5644204A), cited 95 times, followed by a 2003 Japanese attitude-control patent cited 37 times. These older filings function as foundational prior art that later applicants build on and cite around, rather than as the current commercial leaders. For present-day filing activity, China accounts for the largest share of receiving-office filings in the dataset at 28, ahead of the United States at 18 and the EPO at 13, so recent claim density is concentrated more in Chinese filings than citation counts alone suggest.

Filing activity rose from zero in 2017 to a peak of 26 families in 2023, then appears to ease off through 2024-2026. Because publication typically lags filing by around 18 months, the most recent one to two years in any patent trend will always look artificially low, so the apparent 2025-2026 slowdown should not yet be read as a genuine drop in R&D activity. The clearer signal is that growth has flattened rather than continuing to accelerate after the 2023 peak, which is worth revisiting once later filings finish publishing.

Three IPC subclasses carry almost all of the claim density: B62D (motor vehicles and steering) at 55 records, B25J (manipulators and robots) at 47, and G05D (control of non-electric variables) at 27. B62D outranking B25J is notable because it means locomotion and balance-control claims are frequently filed under vehicle/steering mechanics rather than robotics classification, which a search scoped only to robot-specific IPC codes would miss. G05B (control and regulating systems generally) adds a further 14 records, and smaller counts appear in F16F (springs and dampers), G06F, A63H and G06N.

Whole-body control refers to coordinating a legged robot's full set of joints and limbs as a single dynamic system, rather than controlling each leg independently, typically using a whole-body dynamic model alongside models for foot-end contact and collision. It matters commercially because claims in this area, such as the representative 2026 filing from Tsinghua University, tie specific control steps to specific models (foot-end dynamic model, discrete collision model), which narrows what exactly is protected and what a competing implementation using a different model structure might avoid infringing.

Based on the IPC composition, sub-areas tied to AI-based gait prediction (G06N overlap, only 2 records) and wheel-legged hybrid transition control remain thin relative to the core locomotion and balance-control claim base. Impact-aware swing-leg trajectory planning and foot-end collision modelling appear in recent filings like the 2026 Tsinghua application but are not yet densely claimed across the corpus. Before filing into any of these branches, a targeted search against the B62D and B25J classification split described above is worth running, since claim density in the adjacent vehicle-steering classification can be easy to miss.