PROBABILISTIC ROBOTICS

Detecting and Tracking of Dynamic Objects Using RGB-D camera and Computer Vision Applied to Probabilistic Robotics

In the field of mobile robots, a major obstacle is detecting and tracking moving objects (DATMO). This is of particular interest for autonomous robotic systems that do not just need to detect obstacles but to identify, track, and react based on what objects they are interacting with. This research presents a primitive approach to detect and track dynamic objects in an indoor environment for a static mobile robot that uses a particle filter and computer vision to estimate the state of the dynamic objects through a probabilistic approach. In addition, the research lays a foundation to assist in autonomous navigation in self-driving vehicles as detecting and tracking objects are essential in order for the vehicle to make safe decisions and traverse through busy environments.