Whole-body control coordinates every joint of a complex robot at once to juggle many goals — balance, reach, and contact — turning a humanoid's 30-plus joints into a single, unified control problem solved in real time.
Whole-body control moves all of a robot's joints together to do several things at once — stay balanced, reach for something, keep its feet planted — instead of controlling one limb at a time. It's how a humanoid does everything in coordination.
🎯 Quick challenge
Whole-body control is needed because a humanoid must…
A humanoid reaching for a high shelf must, at the same instant, keep its balance, keep its feet planted, avoid hitting itself, and actually reach the target. You can't solve those one limb at a time — they interact. Whole-body control solves them together.
The problem
A humanoid has 30+ joints and must satisfy many objectives and constraints at once:
Tasks — reach the target, look at something, keep the center of mass balanced.
Controlling the arm alone would tip the robot; balancing alone would fail to reach. Whole-body control treats the entire robot as one system and finds joint commands that serve all goals simultaneously.
Many goals, one solve
All objectives and constraints go into one optimization each control cycle; the solution is a single set of joint torques that best serves them together.
How it works
Whole-body controllers typically pose each control cycle as a prioritized optimization (often a quadratic program) over the robot's equations of motion. High-priority objectives (don't fall, honor contacts) are satisfied first; lower-priority ones (reach, gaze) are met in the remaining freedom — exploiting the robot's redundancy, the extra degrees of freedom beyond what any single task needs. It generalizes inverse kinematics from "place one hand" to "satisfy everything at once, including dynamics and forces."
Where you'll see it
The balance and manipulation controllers on humanoids (Atlas, Digit, Optimus) and on advanced quadrupeds that must locomote and manipulate. It's what lets these robots do multiple things fluidly — walk while carrying, catch themselves (push recovery) while reaching, step onto uneven ground while staying upright.
Why it matters
Whole-body control is the framework that makes truly capable legged, manipulating robots possible. It's the answer to the defining challenge of high-degree-of-freedom robots: doing many interdependent things at once, safely and in real time.
