Compositional Diffusion Models for Powered Descent Trajectory Generation with Flexible Constraints

TL;DR

This paper presents TrajDiffuser, a diffusion-based model that generates flexible, multi-constraint powered descent trajectories for spacecraft, improving planning stability and reducing training data needs.

Contribution

The introduction of TrajDiffuser, a compositional diffusion model capable of generating diverse trajectories with flexible constraints for 6-DOF powered descent guidance.

Findings

Generates stable, long-horizon trajectories in real-time.

Supports compositional constraints for diverse scenarios.

Enhances optimizer initialization for faster convergence.

Abstract

This work introduces TrajDiffuser, a compositional diffusion-based flexible and concurrent trajectory generator for 6 degrees of freedom powered descent guidance. TrajDiffuser is a statistical model that learns the multi-modal distributions of a dataset of simulated optimal trajectories, each subject to only one or few constraints that may vary for different trajectories. During inference, the trajectory is generated simultaneously over time, providing stable long-horizon planning, and constraints can be composed together, increasing the model's generalizability and decreasing the training data required. The generated trajectory is then used to initialize an optimizer, increasing its robustness and speed.