Quadruped robot solves large Rubik's Cube using precise foot control
Solving a Giant Rubik's Cube with a Quadruped Robot
Robotics
Summary
Solving a Rubik’s Cube requires turning its layers carefully while holding the cube steady. The authors built a four-legged robot that uses only its feet to turn the layers and move the cube around to solve a giant Rubik’s Cube. The robot uses onboard cameras and sensors to see the cube’s position and colors, and it plans its moves carefully to avoid messing up its progress. It learns these movements through training, allowing it to adjust if the cube or robot shifts. The system successfully completed multiple full solves, showing robots can perform complex, multi-step tasks by combining seeing, planning, and moving.
What this means in practice
- •For logistics automation teams: Use foot-based manipulation combined with onboard perception to handle and assemble large objects requiring coordinated multi-step movements.
- •For robotics software developers: Develop whole-body skills for quadruped robots to precisely manipulate articulated objects while maintaining balance and recovering from disturbances.
Authors
Lukas Amiet, Gabriel Sánchez-Blanco Repgen, Mayank Mittal, Victor Klemm, Andrei Cramariuc, Marco Hutter, Cesar Cadena
Abstract
Pedipulating a freestanding articulated object requires controlling its pose and internal configuration while maintaining balance. We demonstrate this capability by autonomously solving a giant Rubik's Cube with a quadruped using only its feet. The task demands repeated precise interactions, since unintended layer motion can invalidate subsequent operations. We develop an onboard RGB-D perception pipeline that estimates the cube's metric pose and its full color state from partial views acquired through locomotion and flipping. A planner converts the computed solve into robot-executable skill sequences. Learned whole-body skills reorient the cube while preserving its layer configuration and turn its layers with one foot pinning the cube as the other pushes. We train these skills using reinforcement learning with privileged observations and distill them into policies that map onboard depth and proprioception to joint targets. The deployed policies tolerate variations in relative robot-cube placement and recover from external disturbances. The system completes three autonomous hardware solves, each about 30 steps over 8-9 minutes. These results show that combining onboard perception with coordinated pedipulation enables a quadruped to complete extended multi-step physical tasks. Additional material can be found at: https://leggedrobotics.github.io/rubik-skills/