Real time motion tracking algorithm for search and rescue robots

Being able to provide timely help to victims trapped in various disasters is very important. In this paper a novel strategy for tracking and following of victims using unmanned ground vehicle (UGV) is discussed. The proposed system is based on motion tracking algorithm and is useful in operations like fire fighting. The system separates the moving objects from the background. Hence further processing can be achieved through a simplistic method to achieve a real time result. The conventional systems are operator dependent and suffer from the disadvantages of causing delays through wireless communication. The proposed system overcomes these disadvantages. The system uses an ego-motion compensation algorithm to identify motion and a particle filter to track the moving object. Further, in this work the existing ego-motion compensation algorithm is modified since it can not be used along with a continuous control strategy. This is because it is unable to autonomously adopt to variable robot velocity. Such an ability is essential for search and rescue robots. Hence, a velocity based feedback is added to the existing ego-motion compensation algorithm. The effectiveness of this modification in subtracting the background more accurately than other existing algorithms for various velocities is tested using a UGV.