TANGO: Whole-Body VLA Enables Humanoid Robots to Navigate Cluttered Spaces

TANGO, a CoRL paper, presents a whole-body VLA framework that predicts 29-DoF joint actions from natural language instructions and egocentric RGB input, letting humanoids navigate cluttered indoor environments zero-shot using only simulation data.

2026-09-09 ~ 2026-09-09 · 2 related posts

• TANGO: whole-body VLA model navigates humanoid robots in cluttered spaces from sim data — Anqi Li · 2026-09-09
• TANGO: whole-body VLA lets humanoid robots zero-shot navigate cluttered spaces on Unitree G1 — chris_j_paxton · 2026-09-09