MiTaS multi-resolution tactile imitation learning hits 80% success vs 35.5% vision-only baseline
Jan_R_Peters · x · 2026-10-23
The PEARL lab at TU Darmstadt and collaborators released the final version of MiTaS (Multi-Resolution Tactile Sensing), to be demoed live at CoRL 2026 in Austin, Nov 9–10.
- Fuses heterogeneous tactile sensors at different temporal resolutions — RGB camera, vision-based GelSight Mini, and high-frequency event-based Evetac — via modality-specific convolutional stems and transformer-based fusion;
- The multisensory representation conditions a flow-matching policy for contact-rich manipulation;
- Across five contact-rich tasks, MiTaS averages 80% success, vs 35.5% vision-only and 55.5%/63% visuotactile baselines;
- Paper on arXiv (2606.06281), project page and code public.
More from Embodied
- Dev builds ESP32 desk display showing Claude Code and Codex agent status via local Go helper — tonylook · 2026-10-23
- Dream Labs emerges from stealth with Dream-1 robot foundation model at $3/hour — scott_e_reed · 2026-10-23
- AdaDex turns seconds of human hand motion into generalizable dexterous robot skills — mangahomanga · 2026-10-23
- Perplexity's Portable Computer lands on AMD Ryzen AI Max, running local 27B model — AravSrinivas · 2026-10-23
- MIT CSAIL's X-Hinges 3D-Prints Everyday Objects With Built-In Sensing Across Three Motion Axes — MIT News AI · 2026-10-23
- Formic raises $50M Series B to scale physical AI across 100+ US factories — garrytan · 2026-10-23