Belangrijkste concepten
Optimal control solutions for tracking distributions in Wasserstein space can be achieved through a model predictive control scheme, enabling real-time causal tracking of unknown references.
Samenvatting
The content discusses the problem of optimal swarm tracking by formulating it as a distribution tracking issue in Wasserstein metric. It introduces a model predictive control scheme to achieve causal tracking control of a priori-unknown references. The paper is structured into sections covering Introduction, Background and Problem Formulation, Noncausal Solutions Structure, Model-Predictive Control Scheme, and Simulations. Various classes of demand signals are simulated to demonstrate the effectiveness of the proposed control algorithm.
Introduction:
Swarm applications in diverse settings.
Challenges in planning and coordinating motion for large swarms.
Importance of modeling large-scale swarms as distributions.
Background and Problem Formulation:
Notation and preliminaries related to optimal mass transport concepts.
Problem setting with resource and demand distributions dynamics.
Objectives of optimal tracking control problem balancing distance-to-demand and efficiency-of-movement costs.
Noncausal Solutions Structure:
Necessary conditions for optimality described.
Discussion on noncausality acceptability based on deterministic or predetermined demand scenarios.
Model-Predictive Control Scheme:
Description of MPC scheme using noncausal solutions with predictive demand model.
Four-step process involving prediction, optimal control computation, application over short horizon, and state update.
Simulations:
Constant Demand: Resource particles track weighted average of three Gaussians.
Fading Demand: Resource particles shift positions as demand changes over time.
Constant-Velocity Demand: Resource lags behind constant velocity demand components.
Periodic Demand: Resource particles converge to circular path following periodic demand trajectory.
Statistieken
この問題に対する最適なスウォームトラッキングの解決策は、モデル予測制御スキームを介して達成されます。