Research map / Probabilistic and scientific modeling / Dynamical systems
Adaptive Neural Dynamics for Robust Geometric LiDAR-Inertial State Estimation on UAVs
Sitian Peng, Rui Wang
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| replace FAST-LIO2 random-walk bias with neural SDE hybrid bias model | FAST-LIO2 · UAV-Lab | APE (m): 2.13 -> 1.34 | Improved |
| replace deterministic NODE drift with stochastic neural SDE | FAST-LIO2-NODE · UAV-Lab | APE (m): 1.78 -> 1.34 | Improved |
| add adaptive hybrid gate (linear prior mixing) (ablation) | NeuroSDE-LIO · UAV-Lab | APE (m): 1.63 -> 1.34 | Improved |
| add neural dynamics (NODE) to the hybrid model (ablation) | Base+Hybrid · UAV-Lab | APE (m): 1.76 -> 1.52 | Improved |
| add SDE diffusion on top of NODE (ablation) | Base+Hybrid+NODE · UAV-Lab | APE (m): 1.52 -> 1.34 | Improved |
| add hybrid linear prior to FAST-LIO2 (ablation) | FAST-LIO2 · UAV-Lab | APE (m): 2.13 -> 1.76 | Improved |
| use NeuroSDE-LIO bias model | FAST-LIO2 · UAV-Lab | RPE (deg/m): 1.87 -> 1.12 | Improved |
| use NeuroSDE-LIO with differentiable covariance propagation | FAST-LIO2 · UAV-Lab | NIS within 95% CI (%): 70.5 -> 91.8 | Improved |
| use NeuroSDE-LIO bias estimation | FAST-LIO2 · UAV-Lab held-out flight | 6D mean bias L2 error: 0.05329 -> 0.02115 | Improved |
| add neural SDE bias model | FAST-LIO2 · Embedded (Orin NX) | processing time (ms): 45.2 -> 52.1 | Hurt |
| add adaptive mixing gate (ablation) | NeuroSDE-LIO · Embedded (Orin NX) | processing time (ms): 51.6 -> 52.1 | No clear effect |
Closest papers by meaning
- NeRF-LOAM: Neural Implicit Representation for Large-Scale Incremental LiDAR Odometry and Mapping (ICCV 2023 · similarity 0.86)
- EqNIO: Subequivariant Neural Inertial Odometry (ICLR 2025 · similarity 0.86)
- SOAR: Regression-based LiDAR Relocalization for UAVs (NeurIPS 2026 · similarity 0.86)
- Tartan IMU: A Light Foundation Model for Inertial Positioning in Robotics (CVPR 2025 · similarity 0.84)
- Adaptive VIO: Deep Visual-Inertial Odometry with Online Continual Learning (CVPR 2024 · similarity 0.83)
- Markovian Dynamics Enforcer: Feasibility Preserving Correction on Learned Dyanmics Manifolds (NeurIPS 2026 · similarity 0.83)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
