---
title: "JIT Collision Avoidance | Truman DeWalch"
description: "Dissertation research on dust-based just-in-time collision avoidance systems"
url: "https://trumandewalch.dev/research/jit-collision-avoidance"
route: "/research/jit-collision-avoidance"
---

Dissertation Research

# Just-in-Time Collision Avoidance

Dust-based remediation for orbital debris conjunction events

**Timeline:** Sept 2023 – Present

**Advisor:** Dr. Riley Fitzgerald

**Status:** Active — Defense expected Oct 2026

Canister dust meets the primary upstream, hours before the former TCA, so it arrives late and misses the secondary.

## Summary

- Dust-based remediation architecture for debris conjunctions.
- Forward-only propagation with Lambert targeting and event catalogues.
- Uncertainty-aware optimization and sigma-point dispersion modeling.

## Overview

Orbital debris poses an escalating threat to operational spacecraft. With over 34,000 tracked objects and millions of smaller fragments in orbit, the tracked population keeps growing, and operators screen a rising number of conjunctions against it every year. Traditional collision avoidance relies on maneuvering the threatened asset—but many debris objects are uncontrollable, and frequent maneuvers consume precious fuel that shortens mission lifetime.

My dissertation explores an alternative approach: **dust-based just-in-time collision avoidance**. The concept deploys a constellation of small "remediation satellites" that monitor predicted conjunctions. When a collision is imminent, a deployer intercepts the debris *before* the conjunction and releases a targeted dust cloud. The particles impart a gentle momentum change to the debris, nudging it onto a slightly different trajectory—enough to increase miss distance and reduce collision probability without creating additional trackable debris.

## Research Questions

- **Constellation Design:** How should deployer satellites be distributed to maximize coverage of the debris population while minimizing constellation size and delta-v budgets?
- **Dust Dispersion:** How do deployed particles disperse under differential drag, solar radiation pressure, and gravitational perturbations? What deployment strategies maximize momentum transfer efficiency?
- **Uncertainty Propagation:** How does uncertainty in atmospheric density, debris state, and deployment conditions affect conjunction predictions and remediation success rates?

## Methodology

The evaluation pipeline uses a **forward-only propagation** architecture optimized for constellation design studies. Conjunction target states are pre-computed into interpolated event catalogues, eliminating expensive repeated propagation during optimization. The constellation is propagated once per event to an intercept epoch (typically 2 hours before conjunction), where a Lambert solver finds feasible transfers and computes required dust mass.

- **Transfer Solver** Multi-revolution Lambert targeting with configurable policy constraints (max delta-v, minimum lead time, perigee guards)
- **Dust Physics** Sigma-point propagation with Gaussian mixture splitting for accurate uncertainty quantification
- **Event Catalogue** 1,000-event LEO conjunction bank generated via PCA-KDE debris surrogate sampling
- **Optimization** Multi-objective optimization (R-NSDE, NSGA-II variants) with adaptive evaluation policies for computational efficiency

## Current Evaluation Focus

01

### Prototype work

Dispersion-model comparisons in ongoing dissertation research

02

### Ongoing study

Dust-dispersion evaluation under richer uncertainty propagation

03

### 1,000 events

LEO conjunction events in reproducible test bank

04

### Rust-backed

Batch evaluation workflow for optimization studies

## Publications

- *"Stochastic Optimization Techniques for the Design of Just-In-Time Collision Avoidance Constellations"* AIAA SciTech 2026

  Presents the multi-objective optimization framework for designing deployer constellations, including the V3 forward-only evaluation pipeline and adaptive stopping policies.
- *"Dispersion of Targeted Orbital Dust Clouds: Applications to Just-in-time Collision Avoidance"* AAS/AIAA Astrodynamics Specialist Conference 2024

  Details the dust dispersion modeling approach using sigma-point (unscented) transformation for uncertainty propagation and documents the current comparison workflow against simpler baselines.

## Tools & Technologies

Python 3.14 · Rust · C++ · NumPy/SciPy · R-NSDE · UKF · Monte Carlo · STK

[← Back to selected work](https://trumandewalch.dev/#projects)