Rpi with camera
Plot for face-count vs time
Remote client and Server

REMOTE RANGERS

Team Members- Omkar Purandare, Shreyas Vasanthkumar, Virag Gada, Pavan Dhareshwar

Motivation

Many border patrolling units make use of motion detecting sensors and camera to monitor borders for human presence in sensitive areas. Thinking along these lines, developing an application that would make use of the processing power available on the edge to reduce the load on the link and eliminate heavy processing on the server end to serve a needy cause.

Introduction

The Freewave Zumilink Z9-P has support for both RF communication and on-board processing. Since our application is a remote surveillance-based system and many of such systems involve image processing or video streaming to detect motion in the concerned areas we decided to indicate the number of faces present in an image to monitor human presence. This data is transmitted over RF channel to a central server and then uploaded to the cloud through Amazon Web Service.

Description

The aim of the project is to demonstrate the processing capability of a remote radio which reduces the load on the main server and also reduces the data to be transmitted over RF link. For this demonstration, the data is an image which is fed in serially over RS232 to the remote platform and an algorithm is performed on it using openCV to find the number of faces in the image. This processing is performed at the edge on the transmitter and after the data is processed remotely, only important data is transmitted over RF to the server which is a small integer number having the number of faces detected. The server has an internet connectivity and hence after gathering the data, the server uploads the data on the cloud (AWS) in a formatted way with time-stamp.

Final working prototype

The demo consisted of the webcam of the laptop, Freewave Zumilink Z9-P transmitter and receiver, and the display screen of Amazon web services. First, a picture of 3 faces was clicked on the laptop and was serially transmitted to the remote transmitter on which image processing was performed. After the number of faces were detected, the data was transmitted to the receiver and finally uploaded on AWS.

Challenges

1) Figuring out an application due to the limitations with I/O 2) Finding a correct way to send and receive the image file acquired from the camera to the Zumilink board over serial

Learning Outcomes

1) Sending data over RF link 2) Understanding the AWS IoT Embedded SDK and adding code to receive data over RF link and pushing it to the cloud using MQTT 3) Adapting to the available resources and creating an end-end application based project

Folder structure

-----HackCU_RemoteDataProcessing_Zumilink
         |-----remote_client 
                    |----- README.txt
                    |----- main.cpp 
         |
         |-----server
                    |----- subscribe_publish_sample.c
                    |----- aws_iot_config.h
                    |----- Makefile
                    |----- readme.md

Compiling code-
Instructions to compile are inside respective the folders.