2026 Novac Smart Self Mixing Sensor

Self-mixing interferometry is a measurement approach in which a laser beam is re-injected into the emitting laser after self reflection on a target. The objective of this study is to present an updated prototype of an integrated sensor that utilizes self-mixing interferometry embedded with neural networks for enhanced target displacement measurement. We developed a system that incorporates a semiconductor laser functioning as both the emitter and detector, complemented by an embedded platform for real-time data processing. This platform features an Analog-to-Digital Converter (ADC) along with an STM32F75ZIT microcontroller that executes a residual neural network to reconstruct target displacement from interferometric signals. Our methodology included assessing the neural network's resilience against signal amplitude variations and the effects of various quantization strategies on network weights, optimizing the system's overall performance. Results demonstrate the successful real-time reconstruction of target displacement directly managed by the embedded platform, indicating the feasibility of creating robust, low-power, and adaptable sensors using self-mixing interferometry combined with embedded neural networks.