Integration of Vehicle Dynamic Model and System Identification Model for Extending the Navigation Service Under Sensor Failures

1) What problem is this paper solving?

Context: Sensors (GNSS/Camera) fail in extreme conditions (outages, spoofing); VDM drifts quickly.
Core contribution: A sensor-free localization method integrating online System Identification (SI) with VDM.
Achieved goal: Accurate, extended positioning service during complete sensor outages.

2) Why is this paper important?

What changed: Autonomous vehicles need redundancy when primary sensors fail.
Problem created: Standard Vehicle Dynamic Models (VDM) are inaccurate due to modeling errors.
Why current solutions fail: They rely on pre-calibrated parameters that don’t match real-time dynamics.

3) How does this paper solve it?

Contribution 1: Uses online SI to learn real-time powertrain and steering dynamics.
Contribution 2: Feeds identified responses as control inputs to the VDM.
Key result: Reduced mean absolute translation error by 70% compared to conventional VDM.

🎯 Takeaway: Learning the vehicle’s physics in real-time allows it to navigate blindly when sensors fail.

@article{yan2023si,
  author = {Yan, Penggao and Wen, Weisong and Hsu, Li-Ta},
  title = {Integration of Vehicle Dynamic Model and System Identification Model for Extending the Navigation Service Under Sensor Failures},
  journal = {IEEE Transactions on Intelligent Vehicles},
  year = {2023},
  volume = {9},
  number = {1},
  pages = {2236--2248},
  doi = {10.1109/TIV.2023.3273185}
}