Abstract: Heterogeneous air-ground unmanned systems have emerged as pivotal platforms for rapid response in emergency rescue scenarios. However, the significant velocity disparity between aerial and ground units often results in excessive aerial hovering, leading to prohibitive energy consumption and compromised mission endurance. To mitigate this bottleneck, this study develops a collaborative path planning method that integrates high-fidelity environmental modeling with a reverse chronological scheduling strategy, specifically oriented toward system-level energy efficiency optimization. Methods: An integrated air-ground environmental representation is first established by fusing multi-source remote sensing data, including ALOS World 3D-30m (AW3D30) DSM and OpenStreetMap (OSM) vector networks. Three-dimensional voxelized cost maps and travel-time impedance models are implemented to precisely quantify urban obstacles. Subsequently, feasible 3D trajectories are generated using an improved A* algorithm, followed by cubic B-spline smoothing to ensure compliance with non-holonomic kinematic constraints. To resolve the spatio-temporal desynchronization of heterogeneous platforms, an asynchronous scheduling mechanism based on Reverse Chronological Deduction (RCD) is proposed. By designating the Estimated Time of Arrival (ETA) of the ground vehicle as a rigid spatiotemporal boundary, the optimal launch window for the unmanned aerial vehicle (UAV) is determined via backward inference. This approach facilitates a strategic shift from energy-intensive "aerial hovering" to low-power "ground standby." Results: Quantitative evaluations in complex urban environments verify the robustness of the proposed framework in generating collision-free, safe trajectories. Comparative analysis reveals that the RCD-based strategy drastically reduces total system energy consumption from 700.8 kJ (under the ISD strategy) to 239.1 kJ. Conclusions: The proposed methodology achieves a comprehensive energy-saving efficiency of 65.9%. By fundamentally overcoming the energy bottleneck caused by cross-platform speed differences, this research provides a reliable theoretical and technical framework for precise spatio-temporal synchronization in heterogeneous unmanned systems.