Galbot and Tsinghua's LATENT wins IROS award for humanoid tennis forehand

A paper by Galbot, Tsinghua University and partners, nicknamed LATENT, won the Best Entertainment and Amusement Paper Award at IROS 2026. Trained on imperfect motion-capture data from five amateurs, it achieved 96.52% forehand success in simulation and 90.90% on a real Unitree G1.

• IROS 2026 award is for a forehand return, not a full tennis match
• Simulation forehand success 96.52% with 1.32 m average landing error
• On Unitree G1: 90.90% forehand and 77.78% backhand over 20 rallies
• Data came from five amateurs over ~5 hours in a 3x5 m area