AI & ML New Capability

ForceVLA2 introduces explicit force awareness and hybrid control to Vision-Language-Action models, enabling stable contact-rich manipulation.

arXiv · March 17, 2026 · 2603.15169

Yang Li, Zhaxizhuoma, Hongru Jiang, Junjie Xia, Hongquan Zhang, Jinda Du, Yunsong Zhou, Jia Zeng, Ce Hao, Jieji Ren, Qiaojun Yu, Cewu Lu, Yu Qiao, Jiangmiao Pang

The Takeaway

Standard VLAs often fail at physical tasks like pressing or assembling because they only predict positions; ForceVLA2 adds the 'missing link' of force regulation. This allows robots to perform delicate or high-resistance tasks (like wiping or assembling) with significantly higher reliability and safety.

From the abstract

Embodied intelligence for contact-rich manipulation has predominantly relied on position control, while explicit awareness and regulation of interaction forces remain under-explored, limiting stability, precision, and robustness in real-world tasks. We propose ForceVLA2, an end-to-end vision-language-action framework that equips robots with hybrid force-position control and explicit force awareness. ForceVLA2 introduces force-based prompts into the VLM expert to construct force-aware task concep