SoftVTBench: A Deformation-Aware Visuo-Tactile Dataset and Benchmark for Deformable-Object Manipulation
A primary bottleneck is the absence of visuo-tactile datasets that pair policy-visible contact observations with independent physical ground truth over complete tasks. We introduce SoftVTBench, a visuo-tactile dataset for physical-interaction-aware deformable-object manipulation. It contains 4,000 expert demonstrations and more than 50 assets, including volumetric deformable objects and visually matched rigid twins.
Review the primary source, validate the main result, and establish whether any listed-company transmission is direct.
manipulation
hardware
The full methodology, effect size, and limitations still require analyst review.
Company impact remains unverified until a direct economic transmission is established.