Person Following Robotic Suitcase

Abstract

The project aims to build a robotic suitcase capable of vision-based person following using a single camera. A mechanical drive system was designed and attached to a normal suitcase to allow autonomous movement. The robotic suitcase was able to reach 1.7 m/s at no load and 0.9 m/s when loaded to 10kg. Person following algorithm was implemented by tracking the person's shirt based on the color and using PID controller on the centroid position of the red region for heading control and its bounding rectangle contour height for distance control. The robotic suitcase was tested in a straight walking path and a path with a 45 degree turn, 3 different loads and two speed levels. The results show that the developed robotic suitcase performed better at light loads and decreases with heavier loads. It also performed better on the simpler path and during slower walking speed which is caused by wheel slippage at heavy loads.