A common challenge for mobile robots is traversing uncertain environments containing obstacles, rough terrain, or hazards. Without full knowledge of the environment, an unmanned ground vehicle (UGV) navigating towards a goal could easily drive down a path that is blocked (requiring the robot to retrace sections of its path) or run into a hazard causing a catastrophic failure. To address this issue we propose a system of systems (SoS) abstraction to group a distributed set of robots into a single system. Specifically, we propose augmenting the sensing capabilities of a UGV using an unmanned aerial vehicle (UAV). With different dynamic and sensing capabilities, the UAV scouts ahead and proactively updates the plan for the UGV using information discovered about the environment. To predict reachable states of the UGV, the UAV employs a sampling-based method in which a set of virtual particles representing simulated instances of the UGV are used to approximate the distribution of possible trajectories. The UAV assesses if the current UGV path plan is inefficient or unsafe, and if so, provides an alternative path to the UGV. For robustness, a model predictive path integral (MPPI) optimization method is used to modify the waypoints when delivered to the UGV. The strategy is validated in simulation and experimentally.