In this talk my project teammates and I present our work developing a cloud based inference engine to allow embedded applications access to modern object detection systems. For our project we built Chase the object detecting robot using a Raspberry Pi, 3D Printing, and an Amazon Alexa Dot. We show that Chase can find and chase objects around him purely by streaming frames from the front facing camera to the cloud and listening for steering directions sent to him from our detection software. We discuss the implications that this methodolgy has on embedded devices more broadly and show through this proof of concept how one can design and implement an object detecting robot on low-compute power devices.
This presentation was given on January 21st 2018 in the Wu and Chen Auditorium at the University of Pennsylvania in Philadelphia PA as part of the 17th Annual PennApps Hackathon. Our team won 1st place out of 156 teams and won best use of cloud hosting. We each recieved a VR Drone, 24-inch monitor, Amazon fire stick, and a year long subscription to a streaming platform of our choice.