A quiet old grant is a loud constraint on new ones. Boston Dynamics' patent US10081098B1, "Generalized coordinate surrogates for integrated estimation and control" (issued September 25, 2018; inventors Gabriel Nelson and Benjamin Stephens), claims integrated estimation and control for a legged robot — jointly figuring out the machine's state and computing the actions to keep it balanced and moving. It's dated 2018, years before the current humanoid funding wave, and that date is the point.
The classification anchors it in exactly the territory today's humanoid startups are filing in: B25J 9/1666 (manipulator/robot motion planning), B25J 9/1682 (program-controlled robot using state estimation), and B62D 57/032 (vehicles with legs — legged locomotion). Read together, that's the estimation-and-control loop at the heart of any balancing biped. Anyone now claiming a humanoid balance or locomotion-control method is filing into a class where this grant already lives.
Watch the assignment, not the announcement — and watch the prior art, not the keynote. The strategic function of an early, broad, examined control patent is gravitational: it pulls on everyone who comes later. A 2026 humanoid balance claim cannot simply re-describe integrated estimation-and-control of a legged robot; it has to recite something this 2018 grant doesn't, or risk being narrowed or challenged over it. Family geography is roadmap geography, and prior-art geography is constraint geography.
This reframes the humanoid IP race. The new entrants — Figure, Agility, Apptronik and the rest — are not filing onto a blank field. They're filing around incumbents like Boston Dynamics, whose decade-plus head start produced foundational control grants exactly like this one. When a newcomer's balance patent issues, the interesting question for a strategy analyst is what it had to claim narrowly to clear the older art. The 2018 grant is one of the references shaping those narrowing choices.
Caveats, in the desk's voice. Prior-art weight depends on what claim 1 of the 2018 grant actually recites — a broad independent claim constrains widely, a narrow one less so, and you read it to know. Patent term means this grant ages toward expiry, loosening its grip over time. And being prior art is not the same as being asserted; Boston Dynamics holding a foundational method is leverage, not a filed suit. Don't confuse a constraint with a campaign.
For the portfolio-and-litigation beat: when you map the humanoid landscape, date the foundational control grants first, because they set the boundaries the newcomers file within. Boston Dynamics' 2018 estimation-and-control patent is one of those boundary stones. Pull it on PatentBear, read what it actually claims, and use it as the reference point against which the 2026 humanoid balance claims should be measured.