Computer Vision and Autonomous Driving

This project had the main goal of designing and building a robot that could autonomously travel to a series of waypoints positioned in an area while keeping a laser pointed in the positive Y direction, relative to the arena. The arena detailed in the Project 1 Task Specification was 2m by 2m and had a 0.5m by 2m initialization area where teams could set up their robots. A camera vision system consisting of a webcam attached to the ceiling above the arena provided waypoint data to the teams and another camera observed whiteboards set up at the positive Y end of the arena in order to obtain laser pointing data. The vision system observed AprilTags (AR Tags) that were located around the arena: eight AR tags were the border of the arena, four were the waypoints, and an AR Tag carried by the robot was the robot tag.

The vision system also tracked the location of the laser pointer on a whiteboard at the end of the arena. A robot’s score was the root mean squared (RMS) distance between the robot tag’s X position in the arena and the laser’s X position on the whiteboard. A robot was disqualified if its robot tag exits the arena or if its laser pointed in the negative Y direction. Additionally, the robot’s AR tag must always be less than 1m from the floor.

For our final design we went with a "tank" design, a base with a swivel turret on top for the laser. We could manually drive our robot to the first AR tag, but then switch to autonomous mode where it uses the computer vision to drive between the next three AR tags. The biggest challenge in this lab was just to ensure that the robot could turn the exact degree we wanted it to, as well as the laser offsetting the turn simultaneously.