WAMs Combined With Force Data Enhance Pre-training and Fine-tuning

ZhaoMandi · x · 2026-07-03

Researcher Mandi Zhao highlights that incorporating force (haptic/tactile) data into World Action Models (WAMs) simultaneously improves both pre-training and fine-tuning tasks, underscoring the positive impact of multimodal sensory signals on training embodied AI agents.

Original post →

More from Research

Research channel →