arXiv subjects

Minduli Wijayatunga

Publications and source records attributed to Minduli Wijayatunga.

10 recordsLinked to original sources

arXiv2026

Exoplanetary Tour Design with Solar Sails: TheAntipodes Results in the GTOC13 Problem

Solar sails present an attractive but challenging propulsion method for large-scale, long-duration trajectory design problems. In 2025, the 13th Global Trajectory Optimization Competition (GTOC13) presented a trajectory design problem involving an exoplanetary solar sailing spacecraft in the fictional Altaira system, where the goal is to collect scientific return from flybys of planets, comets, and asteroids. High-scoring solutions combine combinatorial gravity assist tour design with continuous solar sail trajectory optimization. This paper presents the solution approach developed by the team `TheAntipodes' during GTOC13. The approach combines several search and optimization stages: (1) trade studies to identify competitive entry opportunities, (2) large-scale beam search over ballistic gravity assist tours to identify beneficial planetary structures, (3) resonant targeting strategies for Vulcan flyby sequences, and (4) multi-leg solar sail trajectory refinement using sequential convex programming (SCP). A key component of the refinement process is the use of a lossless control-convex solar sail formulation, which allows for large portions of the trajectory, including all gravity assist geometry and flyby timing, to be optimized simultaneously to maximize score. The resulting trajectory placed third, with a score of 337.878 from 133 scoring flybys, and exhibited a structure broadly similar to those of the other high-scoring solutions. This demonstrates the scalability of methods such as SCP for very large trajectory design problems.

astro-ph.IM↗

arXiv2026

Memory-Efficient Meta-Reinforcement Learning for Adaptive Safety-Critical Control in Adversarial Spacecraft Proximity Operations

Autonomous spacecraft rendezvous and proximity operations (RPO) require controllers that guarantee safety under thrust constraints while minimizing fuel expenditure. Input-constrained control barrier functions (ICCBFs) provide a control method for nonlinear systems with actuation constraints that construct a forward-invariant safe set. Previous work has shown that learning class-$\mathcal{K}$ functions defining the ICCBF recursion via meta reinforcement learning (meta-RL) yields a robust, non-greedy approach to safety-critical control in RPO. This paper extends that framework further by investigating the performance of three recurrent network architectures (Long Short Term Memory (LSTM), Gated Recurrent Unit (GRU), Selective State Space Model (Mamba)) and two training algorithms (Proximal Policy Optimization (PPO) and Soft Actor Critic (SAC)) to identify the best setup for tuning ICCBF class-K functions via meta-RL. In addition to cooperative test cases, performance is evaluated in the presence of adversarial behavior where the target spacecraft behaves in a way that worsens the safety of the chaser spacecraft. Results indicate that state space models such as Mamba when used with PPO achieve superior task completion, safety, and fuel-savings compared to other architectures, across all cooperative and uncooperative scenarios tested.

cs.LG↗

arXiv2026

Tiny Recursive Models for Solving the J2-Perturbed Lambert Problem

This paper presents a fast, recursive neural solver for the J2-perturbed Lambert problem based on Tiny Recursive Models (TRM), termed the TRM-Perturbed Lambert (TRM-PL) model. TRM is a weight-shared architecture whose effective capacity emerges from iteration depth rather than parameter count: a compact reasoning module is applied repeatedly within a two-level latent hierarchy, refining a candidate departure velocity by simulating the J2 trajectory and correcting it from the resulting tracking error. This unifies initial-guess generation and iterative correction in a single, end-to-end differentiable architecture. The recursive refinement loop is a learned alternative to the homotopy and continuation schemes of classical perturbed-Lambert solvers: rather than following a hand-designed path from the Keplerian to the perturbed solution, the network learns its own sequence of corrections. We evaluate TRM-PL on three test cases of increasing difficulty: single-revolution low-Earth-orbit (LEO) transfers, multi-revolution LEO transfers, and multi-revolution Jovian transfers. Three training paradigms are compared: jointly learning the Lambert solution and the J2 correction; refining the Lambert initial velocity with target-position and J2-corrected velocity supervision; and refining it with target-position supervision alone. Across all cases, the refinement-only approaches are the most reliable. The position-supervised variant reduces the median terminal-position error from 21.7 km to 0.027 km on single-revolution LEO, from 340.9 km to 0.31 km on multi-revolution LEO, all with the same 2.3M-parameter architecture. A single Newton corrector iteration on the TRM-PL output tightens the Jovian median to 0.063 km, yielding compact models accurate enough for embedded deployment.

math.OC↗

arXiv2026

Learning Safety-Guaranteed, Non-Greedy Control Barrier Functions Using Reinforcement Learning

Spacecraft rendezvous and proximity operations (RPO) pose safety risks to high-value assets, so formal safety guarantees are critical. Yet conservative safety controllers can reduce mission efficiency. We propose a unified two-stage reinforcement learning (RL) framework that addresses two complementary limitations of Input-Constrained Control Barrier Functions (ICCBFs) for safety-critical, fuel-limited spacecraft control. Given a certified safe set S, ICCBFs guarantee forward invariance of an inner set C* under input bounds, but the resulting per-step quadratic programme (QP) is greedy and fuel-inefficient within C*, and recoverable states outside C* are conservatively discarded. Stage 1 learns state-dependent class-K-infinity parameters that adapt ICCBF/CLF decay rates, embedding long-horizon cost awareness while preserving invariance in C*. Stage 2 learns a residual barrier h_RL(x) that certifies recoverability for a subset of S minus C*. At run time, the controller selects the appropriate barrier formulation (Stage 1 or Stage 2) and solves a lightweight ZOH QP. Both stages are trained with PPO using rewards that penalise constraint violations, control effort, and task metrics. We evaluate three benchmarks: cruise control, spacecraft rendezvous with a rotating target, and inspection that maximises observability subject to keep-in and keep-out zone constraints. Across test cases, the method reduces median fuel relative to ICCBF baselines by 12 to 25 percent and increases the fraction of trajectories that remain in S by 7 to 8 percent, while retaining real-time QP complexity.

math.OC↗

arXiv2024

GTOC 12: Results from TheAntipodes

We present the solution approach developed by the team `TheAntipodes' during the 12th edition of the Global Trajectory Optimization Competition (GTOC 12). An overview of the approach is as follows: (1) generate asteroid subsets, (2) chain building with beam search, (3) convex low-thrust trajectory optimization, (4) manual refinement of rendezvous times, and (5) optimal solution set selection. The generation of asteroid subsets involves a heuristic process to find sets of asteroids that are likely to permit high-scoring asteroid chains. Asteroid sequences `chains' are built within each subset through a beam search based on Lambert transfers. Low-thrust trajectory optimization involves the use of sequential convex programming (SCP), where a specialized formulation finds the mass-optimal control for each ship's trajectory within seconds. Once a feasible trajectory has been found, the rendezvous times are manually refined with the aid of the control profile from the optimal solution. Each ship's individual solution is then placed into a pool where the feasible set that maximizes the final score is extracted using a genetic algorithm. Our final submitted solution placed fifth with a score of $15,489$.

astro-ph.IM↗

arXiv2023

Convex Optimization-based Model Predictive Control for Active Space Debris Removal Mission Guidance

A convex optimization-based model predictive control (MPC) algorithm for the guidance of active debris removal (ADR) missions is proposed in this work. A high-accuracy reference for the convex optimization is obtained through a split-Edelbaum approach that takes the effects of J2, drag, and eclipses into account. When the spacecraft deviates significantly from the reference trajectory, a new reference is calculated through the same method to reach the target debris. When required, phasing is integrated into the transfer. During the mission, the phase of the spacecraft is adjusted to match that of the target debris at the end of the transfer by introducing intermediate waiting times. The robustness of the guidance scheme is tested in a high-fidelity dynamical model that includes thrust errors and misthrust events. The guidance algorithm performs well without requiring successive convex iterations. Monte-Carlo simulations are conducted to analyze the impact of these thrust uncertainties on the guidance. Simulation results show that the proposed convex-MPC approach can ensure that the spacecraft can reach its target despite significant uncertainties and long-duration misthrust events.

math.OC↗

arXiv2023

Convex Optimization-Based Model Predictive Control for the Guidance of Active Debris Removal Transfers

Active debris removal (ADR) missions have garnered significant interest as means of mitigating collision risks in space. This work proposes a convex optimization-based model predictive control (MPC) approach to provide guidance for such missions. While convex optimization can obtain optimal solutions in polynomial time, it relies on the successive convexification of nonconvex dynamics, leading to inaccuracies. Here, the need for successive convexification is eliminated by using near-linear Generalized Equinoctial Orbital Elements (GEqOE) and by updating the reference trajectory through a new split-Edelbaum approach. The solution accuracy is then measured relative to a high-fidelity dynamics model, showing that the MPC-convex method can generate accurate solutions without iterations.

math.OC↗

arXiv2022

Design and Guidance of a Multi-Active Debris Removal Mission

Space debris have been becoming exceedingly dangerous over the years as the number of objects in orbit continues to rise. Active debris removal (ADR) missions have garnered significant attention as an effective way to mitigate this collision risk. This research focuses on developing a multi-ADR mission that utilizes controlled reentry and deorbiting. The mission comprises two spacecraft: a Servicer that brings debris down to a low altitude and a Shepherd that rendezvous with the debris to later perform a controlled reentry. A preliminary mission design tool (PMDT) is developed to obtain time or fuel optimal trajectories for the proposed mission while taking the effect of $J_2$, drag, eclipses, and duty ratio into account. The PMDT can perform such trajectory optimizations within computational times that are under a minute. Three guidance schemes are also studied, taking the PMDT solution as a reference, to validate the design methodology and provide guidance solutions for this complex mission profile.

math.DS↗

arXiv2022

An Optimised Satellite Constellation for Forest Fire Detection through Edge Computing

The end of 2019 marked a bushfire crisis for Australia that affected more than 100000km2 of land and destroyed more than 2000 houses. Here, we propose a method of in-orbit bushfire detection with high efficiency to prevent a repetition of this disaster. An LEO satellite constellation is first developed through NSGA-II (Nondominated Sorting Genetic Algorithm II), optimising for coverage over Australia. Then edge computing is adopted to run a bushfire detection algorithm using several constellation satellites as edge nodes to reduce fire detection time. A geostationary satellite is used for inter-satellite communications, such that an image taken by a satellite can be distributed among several satellites for processing. The geostationary satellite also maintains a constant link to the ground, so that a bushfire detection can be reported back without any significant delay. Overall, this system is able to detect fires that span more than 5m in length, and can make detections in 1.39s per image processed. This is faster than any currently available bushfire detection method.

math.OC↗

arXiv2022

Team theAntipodes: Solution Methodology for GTOC11

This paper presents the solution approach developed by the team "theAntipodes" for the 11th Global Trajectory Optimization Competition (GTOC11). The approach consists of four main blocks: 1) mothership chain generation, 2) rendezvous table generation, 3) the dispatcher, and 4) the refinement. Blocks 1 and 3 are purely combinatorial optimization problems that select the asteroids to visit and allocate them to the Dyson ring stations. The rendezvous table generation involves interpolating time-optimal transfers to find all transfer opportunities between selected asteroids and the ring stations. The dispatcher uses the data stored in the table and allocates the asteroids to the Dyson ring stations optimally. The refinement ensures each rendezvous trajectory meets the problem accuracy constraints and introduces deep-space maneuvers to the mothership transfers. We provide the details of our solution that, with a score of 5,992, was worth 3rd place.

math.OC↗