What is it about?
We built an AI that uses car vibration sensors to detect potholes, speed bumps, and sudden braking with 96% accuracy, helping drivers stay safe and roads stay maintained.
Featured Image
Why is it important?
Unlike most road-detection systems that rely on cameras (which fail in darkness or rain), our AI uses only vibration sensors—working reliably in any weather. The unique hybrid CNN‑BiLSTM architecture captures both the instant "bump" and the seconds‑long context (e.g., braking vs. pothole), something pure CNN models miss. This delivers 96% accuracy, a compact 298 KB model, and 1.2 ms inference on embedded devices. As cities adopt smart infrastructure and autonomous driving advances, our low‑cost, all‑weather solution enables real‑time road monitoring on smartphones or vehicle OBDII ports—democratizing road safety without expensive cameras or cloud dependence.
Perspectives
What excites me most is how accessible this technology can be. We developed everything in MATLAB, with a model small enough (298 KB) to run on a smartphone or a $50 OBDII dongle. No expensive cameras, no cloud connection, no privacy concerns—just smart signal processing that turns everyday driving into crowdsourced road maintenance. I hope this work inspires researchers to look beyond cameras and high‑compute solutions, remembering that the most impactful technologies are often the simplest, most robust ones. Knowing our algorithm could help a municipality fix a pothole before it damages someone's car—that's the real reward.
Dr. Alejandro Medina Santiago
INAOE
Read the Original
This page is a summary of: Context-Aware Road Event Detection Using Hybrid CNN–BiLSTM Networks, Vehicles, January 2026, MDPI AG,
DOI: 10.3390/vehicles8010004.
You can read the full text:
Contributors
The following have contributed to this page
