LiDAR mapping a rail siding with a sub-250g drone
What a featherweight platform can — and cannot — resolve, with honest notes on where the point cloud fell apart.
Build logs, benchmark write-ups, and field notes. Depth over breadth — real numbers, reproducible methods, honest failure notes.
What a featherweight platform can — and cannot — resolve, with honest notes on where the point cloud fell apart.
INT8 on a Cortex-M55 without losing the tail of the distribution — and what the quantization-aware retrain actually bought us.
A clean-room study on representation choices for condition monitoring — window length, mel bins, and the cost of getting them wrong.
What a featherweight platform can — and cannot — resolve, with honest notes on where the point cloud fell apart.
When the graph optimizer leaves performance on the table, and how to claw it back without a full custom backend.
Latency, watts, and accuracy form a triangle. Picking the corner that ships.