I.C.U

Inspiration

We felt unsafe of our home back in Canada as we came to HackPrinceton.

What it does

Stalls to watch for intruders - once an intruder is detected, the robot will take a picture of the intruder and tweet to its owner. It will then bravely follow the intruder to make sure he does not get away, while insulting at the intruder to inflict emotional damage. When the intruder shows a sign of submission, it will notify the owner once again on twitter with a picture.

How I built it

We mounted Rasberry-pi on the robot for control, using the ROS framework with subscriber model. Intruder detection logic is written in C# with heavy use of Kinect object/human detection. Intruder detection logic runs on a laptop, which is attached on top of the robot. It will then use Ethernet and HTTP to communicate with a server running on the Rasberry-pi, which interfaces with robotic control logic controlling physical parts of the robot. The laptop also sends notification to our web server written in node.js, which handles making API calls to Twitter.

Challenges I ran into

Rasberry-pi being a tough beast, lack of experience with Kinect.

Accomplishments that I'm proud of

We have a whole intruder compliance undertaker (I.C.U.) system that utilizes Kinect, IoT and Web Server, and they work seamlessly together.

What I learned

Solid understanding of Kinetic from ground up,
Architecting communication between logical components
Experience with numerous APIs
Low-level robotic control

What's next for ICUbot

It will be bigger faster stronger to better take down intruders

Built With

Submitted to

HackPrinceton Fall 2015
- Winner Best Microsoft Hack

Created by

I worked on running the mobile robot by reading the data produced by the on board Raspi Flask server. I developed the onboard ROS (Robotic Operating System) nodes that would receive positional data sent from the Kinect to move to robot to chase the human in the Kinect's field of view

Rahul Rawat
I worked on setting up the node server on azure to facilitate sending tweets to the owner based on certain events. In addition, I set up almost the entire kinect component of this project including the human detection and real-time data on the invader's movements.

Mustaqeem Khowaja
I architected data flow between Kinect-hooked laptop and raspberry-pi, wrote server side logic on the pi and co-worked on robotic wheel control to follow intruder based on depth and breadth of target object

Allen Wang
I worked on the node.js server which handles notifying users on twitter when an intruder is detected, as well as the kinect component for human detection.

Darren Cao
Lisa Han