OneBoard – a RaspberryPi and Azure powered IoT solution that converts virtually any flat surface into a shareable digital whiteboard, developed for concurrent collaboration in real-time.
By weighing multiple interface alternatives and rapidly prototyping, I drilled down on the sensor technology that provided a neat balance in mobility, accuracy and speed. I crafted an optimized set of Computer Vision algorithms for resource-constrained RaspberryPi to process the sensor feed at an average framerate of 58 fps – matching the speed of a fast writing user. The most intricate task was to implement concurrent writing, without stifling the real-time experience. I learnt and applied process- based concurrency models in Python, separating computationally heavy implementations from I/O intensive transactions. We designed a minimal data stream (11 bytes-per-frame) that further enabled real-time data transfer to cloud for storage and distribution.
In addition to enterprise collaboration, OneBoard shows promise in expanding higher education to remote villages of India. The work also resulted in a patent application that is under review.