iGPC: Generative Motion Priors for Object-Aware Humanoid Interaction

Key points

• Humanoid robots operating in unstructured environments must combine robust whole-body control with the ability to perceive and physically interact with surrounding objects.1
• While large-scale human motion data provides powerful priors for natural and versatile humanoid control, effectively transferring such priors to perception-driven object interaction remains challenging.1
• To bridge the gap between privileged expert observations and sensory inputs, we propose two complementary training objectives that enable effective adaptation of the pretrained motion prior during distillation.1
• Notably, our experiments across multiple whole-body interaction tasks demonstrate that large-scale generative human motion priors provide an effective foundation for learning deployable policies for humanoid interactions in contact-rich real-world environments.1

Sources (1)

• [1]iGPC: Generative Motion Priors for Object-Aware Humanoid Interaction
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 6, 10:41 AM

  To address this bottleneck, we propose a framework that extends the recently proposed Generative Pretrained Controller (GPC) from general human motion to full-body humanoid-environment interaction.

  Humanoid robots operating in unstructured environments must combine robust whole-body control with the ability to perceive and physically interact with surrounding objects.

Extractive summary: sentences quoted from the sources.