Cooperative Pursuit-Evasion

Overview

Pursuit-evasion problems involve one or more pursuers trying to capture one or more evaders, each acting in their own interest. My research formulates multi-agent nonlinear guidance strategies for cooperative pursuit-evasion scenarios, with theoretically guaranteed conditions for capture (by pursuers) and evasion (by the evader).

The work uses differential game theory and nonlinear guidance principles to derive strategies that are both computationally tractable and formally verifiable. Applications include cooperative interception, active aircraft defence, and autonomous vehicle coordination.

Guidance Formulation

Consider a three-agent planar engagement between a pursuer (\(P\)), an evader (\(E\)), and a defender (\(D\)), each evolving as a nonholonomic agent steered by its lateral acceleration \(a_\ell\):

\[ \dot x_\ell = v_\ell\cos\gamma_\ell,\quad \dot y_\ell = v_\ell\sin\gamma_\ell,\quad \dot\gamma_\ell = \frac{a_\ell}{v_\ell}, \quad \ell\in\{P,E,D\} \]

Classical geometric guidance strategies require the defender, pursuer, and evader to become exactly collinear (a defender–pursuer–evader angle of \(\pi\)) to guarantee interception. My work relaxes this requirement to convergence of the angle \(\chi\) into the wider admissible interval \([\pi/2,\,3\pi/2]\) within a user-specified time, which still guarantees the defender intercepts the pursuer before it reaches the evader — without needing a noisy time-to-go estimate or any assumption on the pursuer's strategy.

Figure not yet added — save pursuit_evasion_geometry.png to images/research/

Key Contributions

• Nonlinear cooperative guidance strategies for guaranteed pursuit-evasion with multiple pursuers.

• Active aircraft defence using cooperative exact-time convergent guidance.

• Geometric guidance for enclosing and tracking moving targets.

• Formal proofs of capture guarantees under realistic vehicle constraints.

• Prescribed-time cooperative geometric guidance for three-agent pursuer-evader-defender engagements, robust to delayed maneuver information exchange.

Relevant Publications