Electronic design project · Jul–Aug 2023
Ultrasonic Localization
Impact localization in solid media
This prototype detects where an impact occurs on an acrylic plate. PZT sensors at the plate edges capture the resulting acoustic response, allowing signal processing and learned features to estimate the impact region.
Technologies: PZT sensors, Raspberry Pi, Raspberry Pi Pico, Python, Keras/TensorFlow
Sensing and acquisition
I built the acquisition and control path around a Raspberry Pi and a Raspberry Pi Pico interface. The circuit connects the plate-mounted PZT sensors to the measurement system, while the prototype provides a repeatable surface for impact tests.
Two localization approaches
I implemented peak-detection preprocessing and compared geometry-based localization with a five-layer multilayer perceptron in Keras/TensorFlow. The geometric method uses the structure of the arrival signals, while the MLP learns features for impact-position classification.
Outcome
The project received a Second-Class Prize in the Guangdong Division of the TI Cup National Undergraduate Electronic Design Contest.