Strategic Evasion Using a Multi-Hypothesis Control Barrier Function

S. Shah, A. Von Moll, and I. Weintraub

Published in Aviation Forum (Submitted for Review), 2026

This paper focuses on the problem of safe, strategic motion planning for an evader in a dynamic pursuit-evasion game. While Control Barrier Functions (CBFs) are a popular method to maintain safety, traditional formulations often rely on fixed, conservative safety margins and isotropic, non-strategic safety constraints. This results in controllers that are either overly conservative or fail to exploit tactical opportunities for evasion. We propose a novel, unified framework that addresses these limitations through two key contributions. First, we demonstrate the advantages of CBF functionality based on variable-speed nonholonomic agents. Second, we construct a set of directional CBFs that create asymmetric safe regions tailored to a pursuer’s straight, left, and right projected maneuvers. To ensure controller stability, we apply a smooth switching methodology that guarantees continuous control inputs during transitions between evasion strategies. The entire framework is unified within a single Quadratic Program (QP) that minimally alters a nominal evasion strategy while satisfying all safety and agent constraints. Through a series of comparative simulations, we demonstrate that our approach significantly enhances evasion performance, enabling successful escapes in scenarios where traditional CBF methods would tend to fail.