In this paper, a new approach is developed for contributing in solving the problem of autonomous mobile robot navigation in unknown environments. This approach is built upon combining fuzzy reasoning and virtual obstacle algorithm to overcome the local minimum problem encountered in presence of concave obstacles by efficiently coordinating priorities between multiple reactive behaviors such as goal reaching, obstacles avoiding, wall following and emergency situations preventing. To achieve this objective, an array of ultrasonic sensors is mounted on the mobile robot providing the distance information between the robot and obstacles. This distance information is used by the virtual obstacle algorithm to calculate some sub-goals for determining the good motion direction to avoid robot trap in local region (emergency situation), since the fuzzy reasoning is used for behavior control of the mobile robot. All the reactive behaviors are mapped into one universe of discourse to guarantee a smooth transition between them especially when the robot moves through closely spaced obstacles. In this manner, the robot oscillations are significantly reduced. Some simulation results are presented to show the ability of the developed approach in performing successfully in complex and uncertain environments.