🤖 AI Summary
In tight formation flight of lightweight UAVs, insufficient modeling of relative pose measurement noise leads to instability, large overshoot, and slow convergence in conventional rigid-graph-based formation control. To address this, this paper proposes a noise-aware adaptive gradient-decomposition formation control method: it pioneers the decomposition of gradient-descent control commands into physically interpretable components—namely, rigidity-preservation, connectivity-maintenance, and noise-suppression terms—and dynamically reweights them based on real-time estimates of relative localization error distributions. The method integrates graph rigidity theory, distributed coordination mechanisms, and stochastic noise modeling to achieve robust closed-loop control. Experimental results demonstrate that, compared to the baseline gradient-based approach, the proposed method reduces system oscillation by 62%, decreases steady-state formation error by 57%, and shortens convergence time by 41%, significantly enhancing control reliability and convergence performance under noisy conditions.
📝 Abstract
A technique that allows a formation-enforcing control (FEC) derived from graph rigidity theory to interface with a realistic relative localization system onboard lightweight Unmanned Aerial Vehicles (UAVs) is proposed in this paper. The proposed methodology enables reliable real-world deployment of UAVs in tight formation using real relative localization systems burdened by non-negligible sensory noise, which is typically not fully taken into account in FEC algorithms. The proposed solution is based on decomposition of the gradient descent-based FEC command into interpretable elements, and then modifying these individually based on the estimated distribution of sensory noise, such that the resulting action limits the probability of overshooting the desired formation. The behavior of the system has been analyzed and the practicality of the proposed solution has been compared to pure gradient-descent in real-world experiments where it presented significantly better performance in terms of oscillations, deviation from the desired state and convergence time.