Voice Activated Object Tracking Quadcopter.

Abstract

The VAOTQ intends to provide a platform for enthusiasts and professionals alike to build on top of an open-source, open-spec system, which composes of two simultaneously running systems: a server, and a client. The server is based on Python and uses YOLO along with Deep SORT for object tracking. Flask is used to create WebSocket APIs which reads a stream of video, passes it through the ML models, and returns the output via a stream of commands. The client end comprises a quadcopter, that has a Raspberry Pi attached to a Pixhawk flight controller. Raspberry Pi, reads the stream from the connected video source and streams it over to the server. Another stream of commands is maintained which is updated in response to the ML model’s output