Many-vs-Many Missile Guidance via Virtual Targets

Marc Schneider, Walter Fichter · 2025

Read Many-vs-Many Missile Guidance via Virtual Targets by Marc Schneider, Walter Fichter (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2511-02526.txt.

Introduction

Marc Schneider and Walter Fichter offered a method for guiding many interceptor missiles against many targets at once, but Schneider later withdrew the submission from arXiv. The withdrawal notice states that later checks showed the proposed method does not generalise beyond the specific scenario considered in the manuscript. The simulation results in the paper should therefore be read as limited to that scenario, not as evidence for the method in general.

As submitted, the work had replaced direct assignment of interceptors to targets with virtual targets. These were predicted trajectory distributions produced by a normalizing-flow model, with each interceptor steered first by zero-effort-miss guidance and then by proportional navigation. Monte Carlo simulations had reported interception gains that grew when interceptors outnumbered targets, the gains the later checks failed to support beyond the tested setup.

Catalog record

CREATORMarc Schneider, Walter Fichter
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE29 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2511-02526

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Many-vs-Many Missile Guidance via Virtual Targets Marc Schneider, Walter Fichter Abstract 1 Introduction The problem of guiding multiple missiles against multiple maneuvering targets represents one of the most challenging scenarios in modern missile defense systems. Many-vs-many missile guidance involves complex decision-making processes including target assignment, energy management, and trajectory optimization, all while accounting for the uncertain future behavior of maneuvering targets. A critical challenge in guidance is energy management, which becomes evident in the midcourse phase of the missile flight. In this phase, missiles must balance the need to maintain sufficient energy for terminal phase maneuvers while also ensuring that they can effectively intercept their targets. Reacting to every maneuver of the target can lead to excessive energy expenditure, especially when the target performs maneuvers that are not aligned with the missile’s trajectory. Not acknowledging any target maneuvers, on the other hand, can result in missed interceptions, particularly when the target performs significant maneuvers that deviate from its predicted trajectory. In many-vs-many scenarios, the problem is further complicated by the need to cooperatively manage multiple interceptors, but this also gives rise to the opportunity to exploit numerical superiority. When multiple missiles are available, they can be distributed across various predicted target trajectories, allowing for a more robust engagement strategy that increases the probability of successful interception. Practically, the problem we address is the following. In a two-dimensional engagement with m maneuvering targets and n interceptors, where the target’s future motion is uncertain, the goal is to maximize the number of successful interceptions. The guidance task is to provide a centralized strategy that distributes the available interceptors across the targets by leveraging stochastic trajectory predictions to maximize the overall and generates midcourse. This requires an effective use of the target prediction and the generation of suitable midcourse and terminal control commands that balance energy use and interception probability. The central difficulty is representing target uncertainty in a compact, actionable form so that n interceptors can be assigned to representative trajectory hypotheses without relying on a large set of hand-tuned maneuver options. Recent advances in missile guidance have explored several complementary strategies to handle target uncertainty and exploit numerical superiority. One approach to handle target uncertainty is the multi-hypothesis method used in one-vs-one engagements. In this approach a set of discrete future maneuvers (for example: left, straight, right) is assumed for the target and guidance is designed to hedge across those hypotheses [ 1 ] . Such multi - hypothesis schemes can improve interception probability by explicitly accounting for maneuver uncertainty, but they tend to be conservative: mitigating a worst - case hypothesis often increases energy expenditure and can reduce overall performance when the target does not execute the assumed worst - case maneuver. Another class of methods considers many - vs - one scenarios. For example, encirclement strategies deploy multiple interceptors to surround a single target and thereby raise the likelihood of a successful hit [ 2 ] . These methods typically assume missiles with controllable speed (e.g., cruise missiles) so that distinct approach geometries and timings can be enforced. Many - vs - many engagements have been addressed using geometric midcourse techniques such as Earliest Intercept Geometry, which derive guidance laws from encounter geometry and timing considerations [ 3 ] . These methods perform well for planar encounters but are typically formulated for two - dimensional problems and do not fully exploit the flexibility of distributing multiple interceptors across a range of plausible target behaviors. The concept of virtual targets (VTs) offers an alternative way to organize many - vs - many engagements by decoupling interceptor guidance from direct pursuit of physical targets. In some works, VTs are implemented as stationary intermediate waypoints that guide missiles until they select a physical target upon arrival [ 4 ] . In other contexts, VTs appear in one - vs - one settings as a sequence of static waypoints that augment otherwise simple guidance laws [ 5 ] . Extending the concept of multiple hypotheses and VTs, a probabilistic trajectory prediction approach presented in our previous work [ 6 ] represents future target positions as a distribution rather than propagating a small set of hand - picked maneuvers. By using a Normalizing Flows (NFs) model to sample from the learned distribution and then time - series clustering those samples, one can construct exactly n VT trajectories that compactly capture the uncertainty across the m maneuvering targets. These data - driven VTs avoid the combinatorial explosion of hand - tuned hypotheses and produce representative, learned trajectory hypotheses that are directly usable by midcourse guidance. It is important to note that the term “virtual targets” in the works [ 5 , 4 ] refers to stationary waypoints, whereas in [ 6 ] the term refers to predicted future trajectories of maneuvering targets. In this paper, we propose a novel approach to many-vs-many missile guidance using VTs generated by the method proposed in our previous paper [ 6 ] . This means that the VTs are not stationary waypoints, but rather a mapping of the predicted position of the target over time. These trajectories serve directly as the predicted target trajectories in the guidance law, without the need for intermediate waypoints or switching strategies. Figure 1 illustrates the difference between the VT-based prediction (upper row, Fig. 1(a) – 1(c) ) and a straight-line prediction (lower row, Fig. 1(d) – 1(e) ), which is used as baseline for comparison in this paper. (a) Sample from learned probability density function. (b) Cluster the samples to obtain virtual targets. (c) Use virtual targets as drop-in replacement for predicted target trajectories. [formula] (d) Kinematic model used for straight-line prediction. (e) Straight-line prediction as baseline. Figure 1 : Comparison of the prediction of virtual target trajectories (Fig. 1(a) – 1(c) ) and straight-line predictions (Fig. 1(d) – 1(e) ). Instead of assigning n missiles directly to m physical targets through traditional Weapon Target Assignment (WTA) algorithms [ 7 ] , we propose a centralized strategy that constructs n VT trajectories, each corresponding to a representative future trajectory of the m maneuvering targets based on probabilistic predictions. Each missile is then guided toward its assigned VT using Zero-Effort-Miss (ZEM) [ 8 ] guidance, which takes the predicted future trajectory of the VT into account, and switches to Proportional Navigation (PN) [ 9 ] guidance in the endgame phase. This approach fundamentally addresses the inherent limitations of conventional many-vs-many guidance by treating the problem as a many-vs-distribution scenario, meaning that instead of engaging discrete targets, the system engages a distribution of predicted target positions which are then clustered into VTs. Each missile is guided toward a unique VT trajectory, effectively creating n distinct hypotheses of the targets’ future behavior rather than relying on handpicked maneuvers or a single deterministic predictions for each target. This strategic distribution of missiles across diverse predicted trajectories exploits numerical superiority to significantly increase the probability of successful interception. The primary contribution of this paper is the development and demonstration of this scalable guidance approach using probabilistically generated VTs, validated through Monte Carlo simulations that demonstrate improved interception probabilities compared to conventional approaches, particularly in scenarios with large numbers of interceptors. In the following, we outline the algorithms and

Read the complete text