Sensor fusion design for navigation and control of an autonomous vehicle

This paper presents the design and implementation of a sensor fusion system for navigation and control of an autonomous electric vehicle. The system integrates signals of laser range finders, magnetometers and inertial measurement units (IMUs). The system keeps the vehicle moving on the desired trajectory. Using Kalman filter with fuzzy-logic module, we are able to get the state of the vehicle and decide the behavior of the vehicle by a fuzzy controller. Kalman filter is used commonly to perform data fusion and is considered as the benchmark for sensor data integration. This method requires a dynamic model of global position system (GPS) and IMU errors, a stochastic model of IMU errors, and a priori data of sensor systems. The accuracy of Kalman filter is influenced by stochastic modeling, outage of some sensors, and the divergence that results from approximations during any linearization process and system mismodeling. To improve the performance of Kalman filter, a fuzzy-logic based sensor fusion system is proposed. With the fused sensor information, the vehicle can recognize the desired trajectory and the obstacles in the environment. Combining the sensors' measurement and the control strategy, a navigation and control system of an autonomous electric vehicle is accomplished in this paper.