Propelled lunar drones: a planner that sequences safe, fuel-efficient jumps to map the Moon

Parabolic jumps between waypoints, image segmentation to detect obstacles, and automatic retreat to a safe location: the framework has been tested in a high-fidelity lunar simulation.

Without an atmosphere, it is impossible to fly on the Moon with propellers: future lunar drones will have to move in propulsive jumps, photographing the ground to produce high-resolution terrain models. Two researchers present an autonomous planning framework for these machines.

A global planner sets a series of waypoints; the drone calculates an optimized parabolic jump between each one. To judge whether a landing zone is passable, a specially trained semantic segmentation model identifies obstacles in aerial images, and a quick method estimates which areas are still reachable with the available fuel. If the planned point turns out to be dangerous or out of range, the craft switches to a diversion maneuver to a safe site.

Trajectory tracking is entrusted to a predictive control (MPC). In a realistic lunar simulation environment, the whole produces safe and fuel-efficient jumps, tracked quickly and with low error. The work remains purely simulated at this stage.