Electronics and Cameras for Smart Eyewear
This research project digs into the realm of smart еyеwеar, focusing on the electronics and cameras that constitute integral components of these innovative devices. Developed within the Smart Eyewear Lab (SEL) of the LUXOTTICA joint research center, the first phase of this study is devoted to a comprehensive state-of-the-art analysis. The primary objective is to gain a deep understanding of the smart еyеwеar available on the market. This involves an investigation of the key components, including sensors and hardware components, computing elements, and power supply systems incorporated within existing smart еyеwеar models. The aim is not only to acquire a comprehensive know-how of the components, but also to understand strengths and weaknesses of each device. Indeed, benchmarking plays a central role in this research, in order to enable not only the identification of implementable features, but also the definition of a robust framework for evaluating future design and prototypes. Twenty smart glasses have been selected for the benchmarking activity, and the figures of merit have been defined and categorized into features (hardware attributes), and functionalities (use-cases). The features under investigation span a wide spectrum of aspects, including the mechanical characteristics of the spectacle frames, AR displays, sensors (such as cameras, microphones, IMU, and touchpad input sensor), speakers, power supply units, CPUs, GPUs, and operating systems. For each of the categories, a proposed benchtop setup is proposed. On the other hand, to evaluate the functionalities some user experience (UX) protocols are proposed, such as “Taking a picture”, “Record a Video”, and “Display a Video”.The project also encompasses an exploration of computing platforms (with the help of the computer science team at DEIB), such as Jetson AGX ORIN, Jetson Nano, and Raspberry Pi 3/4 and the camera model employed is the Raspberry Pi CAM 2/3. To meet the tight requirements of computer vision algorithms, such as global shutter, hardware stereo cameras, and wide FOV, image sensors such as AR0234 and OV9821 with M12 lenses have been identified. In the subsequent phases of the project, the goal is to select and develop cameras and other sensors with appropriate computing platforms, selecting combinations that optimize computer vision capabilities. At the end of the project, a DEMO prototype will be designed and fully developed, able to reconstruct ego actions and understand its surroundings through computer vision algorithms.
Back to Alumni