Autonomous vehicles (AV) are known for their ability to perform challenging or risky tasks that would be difficult for humans. In unpredictable environments, AV missions often require real-time path re-planning, adapting to terrain changes, and balancing conflicting objectives like safety, risk assessment, travel time, distance and energy consumption. We have chosen to focus on Unmanned Surface Vehicles (USV) surveillance missions centered around area coverage and using LiDAR (Laser Imaging Detection And Ranging) or camera imagery. To tackle these challenges, we propose an offline/online approach for monitoring missions. It exploits a multi-objective Optimization (MOO) framework. In the offline phase, a set of alternative paths is computed using MOO and archived. The archive is in fact a Pareto front. One of these paths is selected as the initial path for the USV, while the online phase handles dynamic path re-planning in response to encountered obstacles during the mission. The re-planning process efficiently and quickly adapts the drone path by reusing the archived data. Our experiments involve a standard commercial USV and its LiDAR system. Our results demonstrate the benefits of using pre-computed solutions from the offline phase for dynamic path re-planning during the online phase.