This paper deals with the problem of navigation of unmanned vehicles through poorly known environments cluttered with static and dynamic obstacles. The robot is equipped with a LiDAR able to provide a scan of the surroundings and with classical dedicated localization sensors (odometry, IMU). The proposed navigation strategy relies on: (i) a new concept called Enhanced Laser Scan (ELS), which is built from the current laser scan by adding virtual points along the predicted trajectory of the obstacles; (ii) two sensor-based controllers allowing respectively to reach the goal and to avoid obstacles. These controllers relying on the richer information provided by the ESL, they will be able to anticipate and safely avoid both static and moving obstacles; (iii) a high-level decision process allowing a better choice of the sense-of-motion (SOM) around the obstacle and its reassessment if needed.