Ant Group’s Lingbo open-sources LingBot-Touch tactile foundation model

Open-sourcing a tactile foundation model is a meaningful step because touch has lagged vision and language in robot learning, and pooling nearly 5,000 hours of real tactile data across magnetic, vision-based and piezoresistive sensing addresses a fragmentation problem that has kept hardware-specific models siloed. The single cross-hand representation is what makes this more than a data dump: it lowers the barrier for dexterous-hand makers and action-model developers who previously had to build touch stacks themselves. The open question is adoption — whether hardware vendors and robotics labs standardize on this representation or keep proprietary alternatives. Watch for ecosystem uptake rather than benchmark claims.