prof_pic.jpg

Hi, I'm Arunkumar Rathinam

Research Scientist @ University of Luxembourg

I am a Research Scientist at the Interdisciplinary Centre for Security, Reliability and Trust (SnT), University of Luxembourg.

My research lies at the intersection of computer vision and space robotics. I am interested in building AI systems that can bridge the gap between simulation and real data for spacecraft pose estimation, and leveraging these capabilities to enable autonomous relative navigation and on-orbit servicing missions.

At SnT, I lead the ESA-funded DIOSSA project on deep learning–based spacecraft pose estimation, serve as operations lead for the Zero-G Lab, and co-supervise PhD students working on domain adaptation and event-based pose estimation. I also co-organize the AI4SPACE workshop and the SPARK challenge at CVPR.

Prior to joining SnT, I was a Research Associate at the Surrey Space Centre, University of Surrey, where I developed deep learning algorithms for non-cooperative spacecraft pose estimation and contributed to the FAIR-SPACE project. I received my Ph.D. from the University of New South Wales (UNSW) Sydney, where I worked on SLAM-based autonomous navigation for small body exploration. I completed my M.Sc. in Space Science and Technology at the University of Würzburg, Germany, where I contributed to the UWE-4 CubeSat mission.

Scholar ORCID GitHub LinkedIn CV

Recent News

All news →

Selected Publications

Spacecraft Pose Estimation

  1. TALON: Token-Aligned Lightweight Adapters for 6-DoF Spacecraft Pose Estimation

    WACV’27 Accepted

    Abid Ali, Arunkumar Rathinam, Nidhal Eddine Chenni, Djamila Aouada

  2. Efficient Onboard Spacecraft Pose Estimation with Event Cameras and Neuromorphic Hardware

    CVPRW’26

    Arunkumar Rathinam, J. Lecomte, J. Reelsen, G. Lenz, A. von Arnim, Djamila Aouada

  3. SPADES: A Realistic Spacecraft Pose Estimation Dataset using Event Sensing

    SPADES: A Realistic Spacecraft Pose Estimation Dataset using Event Sensing

    ICRA’24

    Arunkumar Rathinam, Haytam Qadadri, Djamila Aouada

    Project Page PDF Code DOI
Domain Adaptation

  1. SCOPE: Saliency-Guided Cross-Domain Alignment for 6D Object Pose Estimation

    WACV’27 Accepted

    Nidhal Eddine Chenni, Arunkumar Rathinam, Djamila Aouada

  2. Domain-Aware Conformal Prediction for Trustworthy Spacecraft Pose Estimation under Sim-to-Real Shift

    iSAIRAS’26 Accepted

    Nidhal Eddine Chenni, Arunkumar Rathinam, Djamila Aouada

  3. Unsupervised Domain Adaptation for Sim-to-Real Object Pose Estimation with Contrastive Alignment and Pseudo-Label Refinement

    IROS’26

    Nidhal Eddine Chenni, Arunkumar Rathinam, Djamila Aouada

Datasets & Benchmarks

  1. Eclipse: A Spacecraft Rendezvous Trajectories Dataset with Controlled In-Orbit Lighting Conditions

    NeurIPS’26 Accepted

    Nidhal Eddine Chenni, Arunkumar Rathinam, Djamila Aouada

  2. SPARK 2024: Datasets for Spacecraft Semantic Segmentation and Spacecraft Trajectory Estimation

    Dataset

    Arunkumar Rathinam, Mohamed Adel Mohamed Ali, Vincent Gaudilliere, Djamila Aouada

    DOI
Detection & Perception
  1. Annotation Free Spacecraft Detection and Segmentation using Vision Language Models

    Annotation Free Spacecraft Detection and Segmentation using Vision Language Models

    ICRA’26

    Samet Hicsonmez, Jose Sosa, Dan Pineau, Inder Pal Singh, Arunkumar Rathinam, Abd El Rahman Shabayek, Djamila Aouada

    Project Page PDF Code
  2. Hybrid Attention for Robust RGB-T Pedestrian Detection in Real-World Conditions

    Hybrid Attention for Robust RGB-T Pedestrian Detection in Real-World Conditions

    RA-L

    Arunkumar Rathinam, Leo Pauly, Wassim Rharbaoui, Anis Kacem, Vincent Gaudillière, Djamila Aouada, et al.

    Project Page PDF Code DOI
  3. 3D-Aware Object Localization using Gaussian Implicit Occupancy Function

    3D-Aware Object Localization using Gaussian Implicit Occupancy Function

    IROS’23

    Vincent Gaudillière, Leo Pauly, Arunkumar Rathinam, Albert Garcia Sanchez, Mohamed Adel Musallam, Djamila Aouada

    Project Page PDF Code DOI

View all publications →

Datasets

Open datasets I have authored for spacecraft perception. Download them from Hugging Face.

SPARK 2026

CVPR 2026 Challenge

Synthetic RGB spacecraft images with bounding boxes and segmentation masks for 10 spacecraft classes (~360k samples), released for Stream 1 of the SPARK 2026 challenge.

Hugging Face Website

ECLIPSE

NeurIPS 2026

Spacecraft rendezvous trajectories with controlled in-orbit lighting conditions.

Hugging Face Paper

SPADES

ICRA 2024

Synthetic and real event-camera data of a satellite for spacecraft pose estimation — 300 synthetic trajectories with 179,400 pose labels.

Hugging Face Paper Website

SPADES-RGB

RGB companion

RGB spacecraft images with 6-DoF pose, bounding-box and keypoint annotations (~180k samples).

Hugging Face

SPARK 2024

CVPR 2024 Challenge

Monocular RGB sequences with per-frame 6-DoF pose labels for spacecraft trajectory estimation — 100 training trajectories (30,000 images).

Hugging Face Paper Zenodo

SPARK 2022

ECCV 2022 Challenge

110,000 RGB images of 10 spacecraft and a debris class with bounding-box labels, for spacecraft detection and classification.

Hugging Face Paper Zenodo

AKM

Zenodo

Textureless space target dataset for pose estimation of a known texture-less target.

Paper Zenodo