Boston Dynamics Went Electric. Its Old Hydraulic Patents Explain Why That Was Hard.

When Boston Dynamics retired its hydraulic Atlas for an all-electric successor, the press framed it as progress. The patent record frames it as a trade — and shows precisely what had to be repaid. Start with the hydraulics the company spent years perfecting.

The B25J and F15B filings tell a different story than the retirement announcement. Grant US10273986B2, "Discretized valve state control for multi-level hydraulic systems" (2019), is a deep claim on controlling hydraulic actuators with fine, fast modulation. Hydraulics give you enormous force in a small package and a natural compliance — the limb can absorb a shock instead of fighting it. That is why the hydraulic Atlas could do backflips: the actuators delivered burst power that early electric motors struggled to match.

Pair that with US10351189B2, "Whole body manipulation on a legged robot using dynamic balance" (2019), and US10456916B2 on robotic step-path determination, and you see the control stack that the force-dense hydraulics enabled: a robot that uses its whole body, including its limbs, to stay balanced while doing work. The hydraulics weren't a detail; they were the substrate the balance control assumed.

Strip the demo lighting and you get the cost. Hydraulics leak, need pumps, run hot, and are maintenance-heavy — fine for a research showcase, hostile to a product you'd deploy at scale. Electric actuators are clean, efficient, and serviceable, but historically couldn't match hydraulic force density or compliant response. Going electric meant rebuilding, in motors and control software, the very properties — burst force, shock absorption — that the hydraulic patents had baked into hardware.

There's even a tell in the failure-handling IP: US10434651B1, bluntly titled "Failure mode," is a hydraulic-era grant on what the robot does when something breaks. A force-dense hydraulic limb that fails is dangerous; the company patented the graceful collapse. An electric stack has to re-earn that safety story from scratch.

What the patents don't claim is that electric is worse — only that it's different, and that the difference is large enough to require a new generation of IP. The boring incumbent here is the hydraulic actuator that already worked. Reading the old filings is the cleanest way to size what the electric Atlas had to rebuild before it could ship.