METRO — Magnetic Road Markings for All-weather Perception

A robust road surface marking system using passive magnetic tags and a low-cost magnetometer array.

📄 Paper (ACM SenSys 2023) 🧑‍💻 Code of METRO

METRO encodes lane markings and transverse road symbols (e.g., arrow markings) using passive magnetic tags and reads them with a vehicle-mounted magnetometer array. Magnetic sensing is robust to visibility degradation (e.g., rain/snow/fog/occlusion) and complements cameras/LiDAR while remaining low-cost and deployable on real roads.

Motivation — a complementary path when vision degrades

Autonomous vehicles heavily rely on cameras and LiDAR to perceive road markings for lane keeping and navigation. However, existing vision pipelines degrade in rain, snow, glare, or occlusion; lane paint can be covered or worn. In contrast, magnetic fields are resilient to occlusion and can be sensed reliably and efficiently, thus improving the robustness of lane and symbol perception.

System at a glance

METRO is a full-stack magnetic road-marking system with two core components:
1) Passive encoding — passive and durable NdFeB magnets embedded/attached near lane paint encode lanes & transverse symbols via polarity and spacing ratios;
2) On-vehicle sensing & decoding — a 1-D automotive-grade magnetometer array under the front bumper samples tri-axial fields at high rate and a lightweight pipeline detects each on-road magnet, recovers distance ratios, and outputs machine-readable lane/symbol semantics.

METRO comprised of highly deployable, all-weather magnetic dots and a novel sensor array.

How METRO works — design & innovations

Encode. Arrange magnets so that polarity and inter-magnet spacing ratios jointly encode lanes and transverse symbols (arrows/text).

Sense. Sample tri-axial fields along the bar; use first-derivative zero-crossings to detect peaks/valleys and infer magnet polarities; align samples using vehicle speed/heading (e.g., CAN bus) to obtain reliable distance ratios.

Denoise. Run AMN — Adaptive Magnetic-field Neutralization with reference sensors near wheel wells; an LMS-style filter cancels wheel-induced periodic noise with lightweight and real-time computation.

Decode. Recover the embedded symbol/lane semantics for vehicle’s guidance and control.

From left to right: encoding scheme; derivative-based peak detection; AMN wheel-noise cancellation.

System implementation

METRO is implemented and deployed using commodity vehicles equipped with a bumper-mounted magnetic sensing bar and reference sensors near the wheel wells for adaptive noise calibration. The sensing bar integrates multiple COTS triaxial magnetometers (MLX90393) using a modular PCB design, connected to an embedded controller (Teensy 4.1) for high-rate sampling and onboard processing. Reference sensors continuously capture wheel-induced magnetic disturbances, which are adaptively canceled through AMN.

For deployment, passive magnetic tags (e.g., cubic NdFeB magnets) are installed adjacent to painted lane markings such as speed limits, divider lines, and stop lines, fully compatible with standard roadwork procedures. A 3D-printed protective shell shields the magnets from environmental wear and tear. This design enables non-invasive installation, low maintenance, and high durability.

Implementation and deployment of METRO: A–B show the sensing array installation on vehicles; C–D show reference sensors for wheel-noise calibration; E–G show real-road tag deployments; H–J show tag geometry and dimensions.

Experimental evaluation — setups and key results

We conducted extensive on-vehicle field tests covering a wide range of vehicle speeds, sensor–ground clearances, and adverse weather/visibility conditions (snow, rain, fog, glare), and compared METRO against a camera baseline on public-road deployments. Across these scenarios, METRO achieves an overall detection/interpretation rate of >96%, validating its practicality and robustness for real-world road marking perception.

Key results (at a glance)

  • All-weather reliability. Stable decoding under snow cover, rain, fog, and glare, where the camera baseline degrades.
  • Speed & clearance tolerance. Distance-ratio decoding remains consistent across typical driving speeds (up to 55 mph) and bumper clearances (up to 35 cm).
  • Noise resilience. AMN (reference sensors + LMS) effectively suppresses wheel-induced periodic noise, enabling lightweight on-vehicle processing.
  • Deployability. Passive tags and COTS magnetometers on a bumper-width modular PCB support low-cost installation and maintenance for long-term real-road scenarios.

Publication

Jike Wang, Shanmu Wang, Yasha Iravantchi, Mingke Wang, Alanson Sample, Kang G. Shin, Xinbing Wang, Chenghu Zhou, and Dongyao Chen. 2024. METRO: Magnetic Road Markings for All-weather, Smart Roads. Proceedings of the 21st ACM Conference on Embedded Networked Sensor Systems (SenSys ’23). Association for Computing Machinery, New York, NY, USA, 280–293. DOI: 10.1145/3625687.3625809

📚 Cite our work (BibTeX) 
@inproceedings{10.1145/3625687.3625809,
  author    = {Wang, Jike and Wang, Shanmu and Iravantchi, Yasha and Wang, Mingke and Sample, Alanson and Shin, Kang G. and Wang, Xinbing and Zhou, Chenghu and Chen, Dongyao},
  title     = {METRO: Magnetic Road Markings for All-weather, Smart Roads},
  year      = {2024},
  isbn      = {9798400704147},
  publisher = {Association for Computing Machinery},
  address   = {New York, NY, USA},
  url       = {https://doi.org/10.1145/3625687.3625809},
  doi       = {10.1145/3625687.3625809},
  booktitle = {Proceedings of the 21st ACM Conference on Embedded Networked Sensor Systems},
  pages     = {280--293},
  numpages  = {14},
  keywords  = {magnetic sensing, all-weather road markings, magnetometer},
  location  = {Istanbul, Turkiye},
  series    = {SenSys '23}
}