Wake up for Touch! Mask-isolated Tactile Alignment Learning in MLLMs
For multimodal LLM developers, Splash enables non-destructive addition of tactile sensing to compact models, solving the zero-sum trade-off between new modality acquisition and preserving existing reasoning.
Splash introduces a mask-isolated tactile alignment learning framework that partitions pretrained parameters into dormant and critical subspaces, updating only the dormant subspace to integrate tactile reasoning into multimodal LLMs without sacrificing vision-language capabilities. It achieves state-of-the-art performance on visuo-tactile benchmarks (SSVTP, TVL, TacQuad) while preserving general-purpose abilities.
Touch supplies the physical grounding needed to perceive intrinsic material properties, such as friction and compliance, that vision alone often cannot resolve. Recent efforts for equipping multimodal LLMs with this tactile sense, however, expose a zero-sum trade-off: the limited parameter budget of compact models forces a choice between acquiring the new sensory modality and preserving the established vision-language reasoning. We present Splash, a mask-isolated tactile alignment learning framework for MLLMs. Splash quantifies the significance of each pretrained parameter, and partitions the parameter space into a dormant and critical subspace. While the frozen critical subspace acts as a stable anchor to safeguard general visual knowledge, Splash updates the isolated dormant subspace to internalize tactile alignment towards LLMs. This selective, non-destructive expansion effectively prevents catastrophic forgetting and ensures non-destructive modality expansion. Extensive experiments show that Splash effectively achieves tactile reasoning without additional inference overhead in the LLM part, demonstrating state-of-the-art performance on visuo-tactile benchmarks, including SSVTP, TVL, and TacQuad, while preserving its original general-purpose capabilities.