One of the most unique & challenging problems I have worked on was during the Rally-to-Tally hackathon by Mitsubishi America. It was organized last year with an inventory management-based problem statement of counting a large number of industrial steel pipes with diameters ranging from 2 to 48 inches in an open warehouse spread over acres of land. Being from a robotics background, I knew that a robot would be the best candidate to do such a job with speed, accuracy & ease. Therefore, S.T.E.W.A.R.D was born.
I know, I know. You are wondering what in the world S.T.E.W.A.R.D stands for? Well,… it stands for Smart TErrestrial WArehouse Robot Deployment.
So I teamed up with my brothers from different technological backgrounds & we started working on the project (well mostly :p).
I started by setting up a simulation environment with life-size steel pipes in Gazebo and the ‘Husky Rover’ from Clearpath Robotics followed by working on the robot’s software stack including a control loop using ROS & Python. During this, one of my teammates worked on making circle detection happen with HoughCircles using OpenCV & python. Once that was complete, Circle Detection was integrated into the robot’s live camera feed using ROS & counting was implemented.
While this was done, my other teammates were working on setting up InfluxDB & Grafana for a visualization dashboard to show Robot’s health stats & mission statistics (counting, pipe diameter, date/time, etc). ROS Topics were linked with InfluxDB using a custom python node to visualize simulated data on the dashboard.
The project ended up winning the First Prize & teaching us a lot!
Watch the Video to see the project in action. Click Here to visit the project repo.