Answers are grounded in this paper and cite their sources.
Coordinated multi-parafoil airdrop requires trajectory tracking, obstacle clearance, and formation keeping under wind disturbances. However, fixed guidance and obstacle-avoidance parameters limit adaptation across flight stages. We propose a safety-aware adaptive pure-pursuit (APP) framework based on twin delayed deep deterministic policy gradient (TD3), termed Safe-TD3-APP. A pretrained agent jointly adjusts the base look-ahead distance, speed-dependent look-ahead gain, and obstacle-avoidance blending distance using tracking, clearance, and formation states. Parameter projection enforces magnitude, update-rate, and prescribed look-ahead-distance design bounds, while a separate risk rule selects preset fallback parameters. Nonlinear six-degrees-of-freedom co-simulations evaluate the framework through controller comparisons, staged ablation, and Monte Carlo trials. In the evaluated comparison scenario, Safe-TD3-APP reduces lateral-offset and formation-error root-mean-square values by 21.8% and 55.3%, respectively, and increases minimum obstacle clearance by 26.06 m relative to active disturbance rejection control. Ablation shows that online parameter adaptation restores positive obstacle clearance, while the complete safety layer further enlarges the safety margin. Monte Carlo trials under initial-condition and gust disturbances yield smaller terminal-position biases and more concentrated landing distributions.
No takes yet. Share an insight, caveat, or question.
Qi et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: