Embedded Systems

Build a vehicle classifier from roadside audio, then test how transformer features, hard-negative mining, and deployment affect accuracy.

Build a NILM project that estimates per-device electricity costs from breaker-level data and learns how machine learning splits household energy use.

Build and test an embedded audio model that separates speech from cafeteria noise, then measure how well it preserves clarity.

Build a smart bandage concept to measure skin impedance, classify healing stage, and practice sensor design, signal conditioning, and embedded inference.

Design a BLE rendezvous protocol and test how mobility prediction can cut battery use while keeping item recovery fast.

Build a thermal governor that keeps swarm robot inference running longer by modeling heat, load balancing, and throughput tradeoffs.

Build a low-cost structured-light scanner and learn triangulation, calibration, and compression while measuring how well it maps real objects.

Build a circuit fingerprinting system that spots tampered components, practices reflectometry, and learns how CNNs classify electrical signatures.

Build an IoT beehive monitor that turns weight, temperature, and sound into early swarm warning signals with Bayesian modeling.

Build and test an RP2040 BLDC controller that identifies motor behavior online and auto-tunes PI gains for smoother, smarter FOC.