Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Optimal Multi-Debris Mission Planning in LEO: A Deep Reinforcement Learning Approach with Co-Elliptic Transfers and Refueling

Agni Bandyopadhyay, Gunther Waxenegger-Wilfing

TL;DR

The paper tackles multi-target active debris removal (ADR) in Low Earth Orbit (LEO) by introducing a unified co-elliptic maneuver framework that integrates Hohmann transfers, safety ellipse proximal approaches, and explicit refueling logic. It compares a Greedy heuristic, Monte Carlo Tree Search (MCTS), and a Masked PPO reinforcement learning agent, all under the same orbital dynamics and resource constraints, demonstrating that the RL approach achieves the best overall balance of mission efficiency and computational feasibility. Across 100 randomized debris fields with a $ ext{ΔV}$ budget of $3$ km/s and a 7-day horizon, the Masked PPO policy visits about $29$–$32$ debris (vs. $15$–$18$ for Greedy and $25$–$29$ for MCTS) while maintaining runtimes of about 1–2 seconds, whereas MCTS can require $10^3$–$10^4$ seconds. These results highlight the potential of modern RL methods for scalable, safe, and resource-efficient ADR planning in realistic space mission settings.

Abstract

This paper addresses the challenge of multi target active debris removal (ADR) in Low Earth Orbit (LEO) by introducing a unified coelliptic maneuver framework that combines Hohmann transfers, safety ellipse proximity operations, and explicit refueling logic. We benchmark three distinct planning algorithms Greedy heuristic, Monte Carlo Tree Search (MCTS), and deep reinforcement learning (RL) using Masked Proximal Policy Optimization (PPO) within a realistic orbital simulation environment featuring randomized debris fields, keep out zones, and delta V constraints. Experimental results over 100 test scenarios demonstrate that Masked PPO achieves superior mission efficiency and computational performance, visiting up to twice as many debris as Greedy and significantly outperforming MCTS in runtime. These findings underscore the promise of modern RL methods for scalable, safe, and resource efficient space mission planning, paving the way for future advancements in ADR autonomy.

Optimal Multi-Debris Mission Planning in LEO: A Deep Reinforcement Learning Approach with Co-Elliptic Transfers and Refueling

TL;DR

The paper tackles multi-target active debris removal (ADR) in Low Earth Orbit (LEO) by introducing a unified co-elliptic maneuver framework that integrates Hohmann transfers, safety ellipse proximal approaches, and explicit refueling logic. It compares a Greedy heuristic, Monte Carlo Tree Search (MCTS), and a Masked PPO reinforcement learning agent, all under the same orbital dynamics and resource constraints, demonstrating that the RL approach achieves the best overall balance of mission efficiency and computational feasibility. Across 100 randomized debris fields with a budget of km/s and a 7-day horizon, the Masked PPO policy visits about – debris (vs. – for Greedy and – for MCTS) while maintaining runtimes of about 1–2 seconds, whereas MCTS can require – seconds. These results highlight the potential of modern RL methods for scalable, safe, and resource-efficient ADR planning in realistic space mission settings.

Abstract

This paper addresses the challenge of multi target active debris removal (ADR) in Low Earth Orbit (LEO) by introducing a unified coelliptic maneuver framework that combines Hohmann transfers, safety ellipse proximity operations, and explicit refueling logic. We benchmark three distinct planning algorithms Greedy heuristic, Monte Carlo Tree Search (MCTS), and deep reinforcement learning (RL) using Masked Proximal Policy Optimization (PPO) within a realistic orbital simulation environment featuring randomized debris fields, keep out zones, and delta V constraints. Experimental results over 100 test scenarios demonstrate that Masked PPO achieves superior mission efficiency and computational performance, visiting up to twice as many debris as Greedy and significantly outperforming MCTS in runtime. These findings underscore the promise of modern RL methods for scalable, safe, and resource efficient space mission planning, paving the way for future advancements in ADR autonomy.
Paper Structure (17 sections, 3 equations, 4 figures)

This paper contains 17 sections, 3 equations, 4 figures.

Table of Contents

  1. Introduction
  2. Related Work
  3. Mission Model
  4. Operational Scenario
  5. Transfer and Maneuver Model
  6. Refueling and $\Delta V$ Budgeting
  7. Mission Constraints
  8. Planning Algorithms
  9. Greedy Heuristic
  10. Monte Carlo Tree Search (MCTS)
  11. Reinforcement Learning Agent
  12. Unified Evaluation Framework
  13. Results and Discussion
  14. Debris Removal Efficiency
  15. Computation Time
  16. ...and 2 more sections

Figures (4)

  • Figure 1: Overview of the co-elliptic Hohmann transfer sequence, showing the initial and target orbits, transfer maneuvers, and the terminal safety ellipse maneuver at rendezvous.
  • Figure 2: Zoomed-in view of the terminal approach phase, highlighting the safety ellipse maneuver for controlled and safe rendezvous with the target debris.
  • Figure 3: Number of debris objects visited per episode by each algorithm across 100 randomized test cases.
  • Figure 4: Computation time (log scale) for each algorithm per test case. Masked PPO and Greedy achieve fast runtimes, while MCTS is orders of magnitude slower.