I’m working on a fault detection system for a small battery-powered motor vehicle using TinyML and wanted to see if anyone here has tackled something similar or is interested in collaborating/exchanging ideas.
The core idea is to use on-device inference to detect anomalies — things like abnormal vibration signatures, current draw spikes, or thermal patterns — and flag faults in real time without needing cloud connectivity. Battery-powered constraint means power budget is tight, so model efficiency is a real design consideration.
I’m an Electrical Engineer by background, so I’m coming at this from the hardware and power systems side rather than purely software — which I think gives an interesting angle on the problem.
Would love to hear from anyone who:
• Has done fault detection or anomaly detection on motor systems with TinyML
• Has experience with current/vibration sensor fusion on constrained hardware
• Is just interested in the problem space and wants to nerd out
Drop a reply or DM me. Always happy to exchange notes.