Suganya Viswanathan, Baskaran Ramachandran, and Deivamani Mallayya
Mobile Robot Navigation, Heterogeneous Sensor Fusion, EKF, Incremental Map building
Mobile Robot Navigation has become almost inseparable component for a robot to behave autonomously in an unknown environment. This paper proposes a methodology for autonomous navigation using heterogeneous sensor fusion primitive. The minimal sensors selected to fusion, with the target of handling both indoor and outdoor environment are accelerometer, gyroscope, magnetometer, 2D camera, radar and GPS. The simulation environment is chosen as an indoor environment with an assumption of less dynamism in the environment. The approach of incremental map building is incorporated to undertake the problems of the dynamic environment. Extended Kalman Filtering is used to localize the moving object and estimate the next pose using the incremental map. This paper insists that the heterogeneous sensor fusion (of Inertial Measurement Unit, Visual sensors, and GPS) assisted autonomous navigation decreases the mean error in the position estimate to (3, -0.5, -1,6) in the NED frame of reference.