Benchmark RadarTacVerse
TacVerse is a multi-sensor dataset and benchmark for cross-sensor vision-based tactile perception, containing 106,800 tactile images from seven vision-based tactile sensors. It supports shape classification, grating classification, and force regression tasks, with evaluation under within-sensor, zero-shot cross-sensor, and few-shot adaptation settings.
- Released
- 2026-06-24
- Readiness
- Paper only
- Primary field
- General AI
Why it matters
TacVerse fills a gap in evaluating generalization across tactile sensor designs, which is crucial for real-world robot deployment. It enables systematic study of sensor shift, data-efficient adaptation, and self-supervised learning in tactile perception, providing a controlled testbed for improving cross-sensor robustness.
Motivation
Vision-based tactile sensors (VBTSs) enable robots to infer contact geometry and force-related cues by imaging deformation through an internal camera, yet generalisation across sensor designs remains poorly understood.
Primary resources
Benchmark Radar records only publicly supported details and links back to primary sources for verification.