Robust Drag-Aware Launch Planning and Shoot-on-the-Move Compensation for a Mobile Robot

A light perforated projectile is difficult to launch consistently from a moving robot because aerodynamic drag changes the flight, launcher commands vary, and the projectile inherits platform velocity. This paper develops and physically evaluates a unified methodology comprising a quadratic-drag simulator, make-window selection, distance-indexed launcher maps, and shoot-on-the-move (SOTM) compensation. A vectorized fourth-order Runge–Kutta solver advances the command lattice; cubic Hermite interpolation and bracketed binary search locate the descending rim event. The deployed rule tests bounded perturbations around each speed–angle command and selects the nominal scoring command with the largest simulated make fraction. Caching the target-independent event map X(v,θ) and querying rectangular windows by prefix sums evaluates the same objective without repeated integration. Because the deployed speed box is wider than the scoring band, the speed choice reduces to a low-speed tie-break that places the simulated crossing 1–21 mm past the front lip, so the physical results also reflect launcher-map calibration. Target entry increased from 70/100 to 93/100 in the development record. In four moving-speed strata, SOTM produced 54/60 entries versus 30/60 without compensation (exact conditional stratified p=1.59×10^-7). Operational records contained 94/97 and 105/110 target entries. The resulting contribution is a reproducible model-to-deployment method with numerical, controlled, and competition-scale physical evidence.