Energy-Efficient Multi-Joint Cooperative Motion Planning for Unitree G1 Humanoid Robot in Complex Dance Scenarios


Yu B., Xiao A., YAHYA H.

3rd International Conference on Machine Intelligence and Digital Applications, MIDA 2026, Virtual, Online, 24 - 26 Nisan 2026, cilt.92, ss.1625-1635, (Tam Metin Bildiri)

  • Yayın Türü: Bildiri / Tam Metin Bildiri
  • Cilt numarası: 92
  • Doi Numarası: 10.3233/atde260449
  • Basıldığı Şehir: Virtual, Online
  • Sayfa Sayıları: ss.1625-1635
  • Anahtar Kelimeler: dynamic modeling, energy optimization, Humanoid robot, inverse kinematics, motion planning, multi-joint coordination, quintic polynomial trajectory
  • İstanbul Gelişim Üniversitesi Adresli: Evet

Özet

Humanoid robots performing complex dance routines face critical challenges in multi-joint coordination, trajectory smoothness, and energy efficiency. This paper proposes a unified mathematical framework integrating spatial analytical geometry, kinematic modeling, and dynamic energy optimization for the Unitree G1 humanoid robot. Firstly, a forward and inverse kinematic model based on homogeneous transformation matrices is established to map joint angles to end-effector positions, yielding a target end-effector coordinate of (-146.35, -84.5, -169) mm for the left arm, with all joint angles verified within motor safety limits. Secondly, a quintic polynomial trajectory planning method combined with an S-shaped velocity profile is employed to generate C2-continuous motions, where the knee joint reaches a maximum angular velocity of 12.34◦/s at 3.45s during non-uniform linear locomotion. Furthermore, a joint dynamics-energy integrated model with an electromechanical efficiency factor (0.85) is constructed to optimize power consumption. Experimental results show the proposed strategy reduces total energy consumption by 15.5% (from 3.10 Wh to 2.62 Wh), with the battery utilization rate lowered to 61.1%. All motion phases achieve over 13% energy saving while maintaining motion fidelity and dynamic stability, effectively enhancing the robot's operational endurance.