An Implementation of Multi-object Tracking Using Omnidirectional Camera for Trash Picking Robot

Trash is unwanted materials that are left after someone has used something or things that has been thrown away. It is needed to manage the trash and offers variety of solutions for recycling items that don't belong to trash. Technological development encourages people to working for trash handling and management. The utilization of technology especially in an implementation of the robot technology will help human to handle and manage the trash which includes trash picking, handling, and classifying activities. One of the problems in trash handling using robot is detecting of the trash as an object and determining the distance from the robot to the trash. The main purpose of the implementation of this system is to provide a new method of trash detection using Omnidirectional camera and distance estimation of the object using Euclidean distance estimation method. This experimental design was conducted on four objects with different colors and the distance with the robot. The result shows that trash picking decision will be executed from the nearest distance first. The using Omni-directional camera as the vision system of the robot is work well, although it had some problem to detect and estimate from the object which is caused by the gradation of light and the object has out of the camera range radius.