In this project we developed a computer vision based method to identify and track common litter found on city side walks. In this project we developed a synthetic dataset that was used to train a Mask R-CNN classifier.
This was a four-wheel drive robot that was designed to follow a human collaborator. Using solely a low-cost lidar and a UWB sensor, this robot used a robotics software package including a custom designed fuzzy-logic controller implemented using Python and ROS to follow its human counter part. This robot could be used to pull cargo, sweep, or cut grass.