IROS 2025

IROS 2025

IEEE/RSJ International Conference on Intelligent Robots and Systems 2025 — Hangzhou, China, October 19–25, 2025.

Scale: 4,306 submissions → 1,991 main-track accepted (46.2%) + 777 RA-L / T-RO / T-ASE / T-MECH journal transfers. Record submission year. Five parallel "Award Finalists" sessions across ~45 papers.

Awards relevant to VLA / manipulation:

• Best Conference Paper — biosyncretic-robot electrodes (Shenyang Institute of Automation) — not VLA-direct
• Best Student Paper — Neural MP (Dalal et al., CMU) — neural motion planner using learned priors
• Best Paper Finalists include VLA/manipulation papers:
  • AgiBot World Colosseo (OpenDriveLab + AgiBot) — 1M+ trajectories across 100 robots
  • Interactive Legged-Manipulator Navigation
  • Unified Whole-Body Policy

Surveys hosted in this wiki

• VLA & Manipulation Survey (IROS 2025) — 18+ IROS 2025 VLA/manipulation papers in 7 categories. Flagship: AgiBot World Colosseo (1M+ trajectories, Best Paper Finalist) + Neural MP (Best Student Paper). Includes a CoRL 2025 → IROS 2025 → NeurIPS 2025 / ICLR 2026 lineage map.

Quick paper index (by category)

VLA architecture (IROS-specific spins)

• RoboNurse-VLA (CUHK MSMRC) — SAM 2 + Llama 2 for surgical scrub-nurse; voice-driven, real-time
• CapsDT (CUHK Ren Lab) — Diffusion-Transformer VLA for magnetically-actuated capsule endoscopy in the stomach
• TinyVLA (Midea + ECNU) — Compact VLA family (40M–1.4B-param backbones); TinyVLA-H beats OpenVLA by +25.7% success at 5.5× fewer params and ~20× lower inference latency
• PD-VLA (HKUST-GZ + PolyU) — Parallel fixed-point iteration reformulation of AR VLA decoding; 2.52× faster, training-free
• ReBot (UNC + UT Austin) — Real-to-sim-to-real video synthesis for VLA adaptation; +20% Franka real for OpenVLA
• Refined Policy Distillation — Distills a VLA into a compact RL expert

Real-robot deployment / systems

• HACTS — Bilateral teleoperation hardware
• Bunny-VisionPro — Vision Pro bimanual teleop
• MoMa-Teleop — Zero-added-cost whole-body mobile-manip teleop
• iDP3 Humanoid — 25-DoF cart-mounted humanoid + 3D diffusion policy

Dexterous manipulation

• VTAO-BiManip — Visuo-Tactile-Action pretraining for bimanual
• LDexMM — Language-conditioned dex multi-task
• RoboDexVLM — Dex manipulation task planning
• DexDiffuser — Diffusion for dex grasping
• FoundationGrasp — Foundation model for grasping

Mobile manipulation

• MORE — Mobile manipulation benchmark
• MoMa-Teleop — also real-robot deployment

Humanoid / whole-body

• Unified Whole-Body Policy — Best Paper Finalist
• iDP3 Humanoid — 25-DoF humanoid + 3D diffusion
• GeT-USE — Whole-body humanoid
• ManiGaussian++ — 3D Gaussian-based manipulation

Data / benchmarks

• AgiBot World Colosseo (Best Paper Finalist, Genie Operator-1) — 1M+ trajectories, 217 tasks, 100 robots, 5 domains
• Neural MP (Best Student Paper) — Neural motion planner, CMU

Teleoperation & imitation-data interfaces

• HACTS, Bunny-VisionPro, MoMa-Teleop — see categories above
• Tactile-VLA (workshop paper, 2507.09160) — tactile + π0-style backbone

Relevant workshops

• Generative AI for Robotics
• Tactile Manipulation (hosts Tactile-VLA best paper)
• AI Meets Autonomy
• RoDGE (Robotic Dexterous Grasping and Execution)
• AI Robot for Science
• Fine Manipulation
• Embodied AI for Science

Trends — what's distinctive about IROS 2025

1. VLAs as vertical-domain products — RoboNurse-VLA (surgical), CapsDT (capsule endoscopy), RoboDexVLM (manipulator planning). IROS is where VLA meets real application with hardware.
2. Real-robot systems integration, not architecture theory. Every paper has a hardware demo. HACTS, Bunny-VisionPro, MoMa-Teleop, iDP3-humanoid all feature hardware novelty as much as algorithm.
3. VLA acceleration, compression, distillation. PD-VLA (2.52× faster), TinyVLA (<1B params), ReBot (real-to-sim-to-real), Refined Policy Distillation — the "how do we actually ship this" papers.
4. Industrial-scale data platforms. AgiBot World Colosseo's 1M+ trajectories dwarfs every prior manipulation dataset.
5. Tactile is a whole research track at IROS. VTAO-BiManip, Tactile-VLA (workshop), Adaptive Visuo-Tactile Fusion, RoTipBot. IROS is where tactile research concentrates.

Cautionary notes

• Harmonic Mobile Manipulation is IROS 2024, not 2025 — commonly miscited.
• IROS papers often have arXiv preprints from late 2024 / early 2025 (RA-L pipeline) and are presented at IROS in October. Chronological lineage claims should account for this.
• IROS has ~40 category-specific awards, not a single Outstanding Paper list — "Best Paper" contests are across multiple tracks.

← Back to IROS · Home