📄 In-Depth Reviews — Per-Paper Long-Forms

Single-paper deep-dives (one page per paper). For the cross-paper topic reviews, lab/series programs, and latest-paper reviews, see the main catalog: In-Depth Reviews.

← Back to In-Depth Reviews catalog · [Home]] · visual map: [Knowledge Graph

---

Architecture / runtime

• Discrete Diffusion VLA · Fast-in-Slow · AsyncVLA · DuoCore-FS · Steerable Policies

Data / training / evaluation

• VLM4VLA · Action Space: EEF vs Joint · RoboMME (memory benchmark)

World models / systems / tactile

• DreamDojo · Genesis GENE-26.5 · OmniVTA (visuo-tactile WM) · FTP-1 (foundation tactile policy)

Hybrid / MoT (three-expert) long-forms

• Motus · HALO · BAGEL · Being-H0.7 · Cortex 2.0 · DYNA-2 · RLDX-1

Dexterous-hand data & interfaces long-forms

• DO As I Do · AnyDexRT · YUBI · DexEXO · T-Rex

Multi-task / in-context / continual long-forms

• MergeVLA · HiMoE-VLA · DyGRO-VLA · ICRT · Behavior Prompting · MimicDroid

IROS 2026 full-paper analyses

• WAM / hybrid: AtomVLA · DreamMimic · Scaling Cross-Embodiment WM
• Policy / RL / efficiency: VLA-RL · IMLE-VLA
• Bimanual: 3D FlowMatch Actor · EquiBim
• In-context / memory: RoboSSM · ICLR: Visual Reasoning · TempoFit
• Dexterous / contact: DexKP-VLA · VCoT-Grasp · MrGrasp · Proprioceptive Transformer · Articulated-Tools Sim2Real · FILIC

RSS 2026 per-paper pages (figure-illustrated)

• 15 figure-illustrated per-paper pages (Ψ₀, π*0.6+RECAP, LDA-1B, mimic-video, ViTacFormer, PolaRiS, OAT, Legato, HoMMI…) are catalogued in the RSS 2026 survey and RSS-2026-Papers index.

---

When adding a new per-paper long-form, list it here; topic reviews go in Reviews. Also note it in Changelog.

← Back to In-Depth Reviews catalog · Home