Extended Kalman Filter (EKF) and K-means clustering approach for state space decomposition of autonomous mobile robots

Path planning and navigation in unknown environment is one of the most challenging tasks for autonomous mobile robots. Decomposition of the state space is vital for avoiding obstacles and generating an efficient trajectory. For the purpose of localization and building an efficient map in an unknown environment, decomposition of this area is equally important. For this reason, an autonomous mobile robot has to manage the free area of its workspace very efficiently. Perfect decomposition of state space will make the map building task very easy, faster and efficient. On the other hand, finding an efficient path is the most difficult part of simultaneous localization and mapping (SLAM) problem. Also, sensors are prone to noise that makes the robot to find the best move really hard. So we integrated the Extended Kalman Filter (EKF) and K-means clustering algorithm to build a method to formulate an efficient estimation of the future states and manage the environment efficiently as well. Here, Extended Kalman Filter (EKF) will find the accurate estimation on sensor data and K-means clustering algorithm will perform the partitioning task on that data to decompose the environment avoiding the obstacles.