Core Concepts
The author presents an enhanced direct-method-based approach to handle path constraints in real-time optimal control problems through mesh refinement. The proposed methodology guarantees constraint violation-free trajectories by constructing trajectory bounds and assessing bound violations.
Abstract
The paper introduces a novel method for real-time optimal control problems with path constraints, focusing on mesh refinement. By extending reachability sets and deriving envelope regions, the approach ensures constraint violation-free trajectories. Numerical simulations demonstrate the effectiveness of the proposed methodology.
Recent advancements in real-time optimal control involve direct methods combined with sequential convex programming (SCP). The direct method partitions the domain into intervals separated by sample points, employing polynomial interpolation to determine state and control variables between nodes. Path constraints are addressed at each mesh point to prevent state variables from entering forbidden regions.
A key focus is on addressing inter-sample collision problems that arise when trajectory segments trespass forbidden regions between mesh points. The proposed method aims to minimize additional mesh insertion while ensuring constraint violation-free trajectories efficiently. By analytically determining trajectory bounds and assessing violations, the approach integrates seamlessly into general direct formulations of optimal control problems.
The study extends existing literature on curvature bounded paths, introducing an envelope concept for reachability sets. By formulating necessary conditions of optimality based on Pontryagin Maximum Principle (PMP), the paper proves convergence of the proposed algorithm and demonstrates computational efficiency through numerical simulations.
Stats
Recent advancements in real-time optimal control involve direct methods combined with sequential convex programming (SCP).
The direct method partitions the domain into intervals separated by sample points.
Path constraints are addressed at each mesh point to prevent state variables from entering forbidden regions.
The proposed methodology guarantees constraint violation-free trajectories through mesh refinement.
Numerical simulations demonstrate the effectiveness of the proposed methodology.