Introduction: Autonomous Snowplow Machine Using AI-generated Object Recognition Skills

My name is Lawrence Park and I am currently a student studying in Fryeburg Academy, Maine. What I made is autonomous snowplow machine using Raspberry Pi, Arduino and 3D Printing. Because I often experience a harsh snowy weather back at my school, I have decided to make a snowplow machine to drive around at my school.

Supplies

  • Raspberry Pi
  • wife dongle
  • Arduino uno
  • RC Car (Needs to be pretty big to go through the snow)
  • Webcam
  • wife router
  • Computer or a Laptop ( any devices that can do the coding)
  • 3D Pen
  • 3D Printing Machine
  • Portable Battery

Step 1: Designing

First, if you want to build the machine, you need to find a goal. My goal for this project was to create the snowplow machine and coding it to clean up the snow in certain area using object recognition skills. I also had to 3D design the drill-shaped block to push the snow away.

Step 2: Breaking Down the RC Car

I used a RC Car as the main body of the machine. I first found the RC Car that I was not using and took the lid off. Then, I cut the wires off from the previous and replaced it with my arduino uno. Now we can control the motors in RC car from my computer.

Step 3: Coding With Raspberry Pi

After putting in the arduino, I had to move on to the next step, which was setting up the machine with Raspberry Pi. First, I had to connect the arduino with my computer using a usb cable. After that, install the raspbian os. Now, open terminal and run sudo apt-get update and sudo apt-get update. Then, install vncserver and Arduino ide. For the purpose of operating the machine remotely, I had to set up a wifi connection with constant ip address. This is when I needed a wifi router to connect to pi. This will be crucial for communication and control.

Step 4: Image Learning and Testing

Before I start the image learning, I had to install a webcam on the main body, using my 3D pen. I decided to put the webcam a little higher than the machine so that way, the main body won't block the webcam vision. After installing the camera, I had to install an OpenCV to start the deep learning. In this model, I utilized deep learning techniques like object detection, which makes the camera ( the snowplow machine) to recognize miniature stop signs. I took multiple pictures of the stop sign and made the machine to deep-learn them. Now the machine knows that when it sees a certain stop sign infront of itself, it will slow down and turn around to its next route. After the deep-learning process, I tested out the program to see if the machine learned a sufficient amount of data.


Attachments

Step 5: Final Step

After these steps, now I have to just test out outside the field and see if its working. After I make sure it clearly works, I have to put on the actual snowplow drill and see if it really works! I also added some additional pictures for you guys so maybe the steps get more clear. Thank you if you like this project!