Alvin Combrink
Portrait of Alvin Combrink
Fig. 01 — Profile picture

Automation / Systems and Control / Department of Electrical Engineering

Alvin Combrink

PhD student on Multi-Agent Path Finding

Chalmers University of Technology

Planning provably safe and optimal motion for high-fidelity MAPF.

Research
My research addresses multi-agent path finding (MAPF) for highly generalised problem variants in continuous time and space. After an early PhD focus on personnel scheduling in healthcare, I pivoted to MAPF, where I develop solvers with formal guarantees; my main contribution being the theoretical restoration of correctness guarantees for exact MAPF in continuous-time. My current work focuses on extending these formal guarantees to practice with an anytime-optimal algorithm (producing an initial solution fast and refining it to optimality) in a formulation general enough to span ground, aerial, and manipulation platforms, and conflicts richer than geometric collision alone.

Research trajectory

Each node is a publication, placed along time and grouped by topic; curves join related results. Hover a node to read it and trace its connections, press for more information.

  • MAPF
  • Personnel Scheduling
  • Motion Planning
  • Network Prediction
Research trajectory graph Publications positioned by date along the horizontal axis and grouped into topic rows, with curves connecting related papers. A full text list of every publication follows below.
Horizontal axis — date  ·  row — topic  ·  hover to read & trace connections, click to open in the list

Publications

13 records

  1. Probabilistically Robust MAPF in Continuous Time

    M. Johannesson, M. Mazen, A. Combrink, S. Roselli, M. Fabian

    Extends the OC-CBS algorithm for probabilistic continuous-time MAPF, where edge traversal times are uncertain, for finding solutions that are robust up to a desired probability.

    TBD 2026 Planned MAPF

  2. Counterfactual Traffic Prediction under Network Reconfiguration - Graph Neural Surrogate for the flow-to-flow problem

    X. Wu, F. Rydin, A. Combrink, B. Kulcsár

    A GNN-based method that predicts post-intervention traffic flows from observed pre-intervention flows and network structure, bypassing the need for OD demand data.

    TBD 2026 In Progress Network Prediction

  3. Zero-Shot Generalization from Motion Demonstrations to New Tasks

    K. Freitag*, A. Combrink*, N. Figueroa

    The Gaussian Graph: combining isolated motion demonstrations into a shared graph structure to enable stable dynamical-system control that generalizes to unseen robotic tasks.

    CASE 2026 In press Motion Planning
    DOI
    BibTeX 
    @misc{freitag2026zeroshotgeneralizationmotiondemonstrations,
 title={Zero-Shot Generalization from Motion Demonstrations to New Tasks},
 author={Kilian Freitag and Alvin Combrink and Nadia Figueroa},
 year={2026},
 eprint={2603.15445},
 archivePrefix={arXiv},
 primaryClass={cs.RO},
 url={https://arxiv.org/abs/2603.15445}}

  4. Anytime-Optimal Continuous-Time Multi-Agent Path Finding

    A. Combrink, S. Roselli, M. Fabian

    Extends the OC-CBS algorithm to an anytime-optimal version for a continuous-time MAPF with heterogeneous agents, providing a practical solution for real-world applications.

    JAIR 2026 In Progress MAPF

  5. A General Formulation for the Teaching Assignment Problem: Computational Analysis Over a Real-World Dataset

    M. Johannesson, L. Brink, A. Combrink, S. Roselli, M. Fabian

    A mathematical formulation of the Teacher Assignment Problem, evaluated with SMT, CP, and MILP solvers on real-world data to produce fairer, more balanced teacher assignments.

    CODIT 2026 In press Personnel Scheduling
    DOI
    BibTeX 
    @misc{johannesson2026generalformulationteachingassignment,
 title={A General Formulation for the Teaching Assignment Problem: Computational Analysis Over a Real-World Dataset},
 author={Moa Johannesson and Lina Brink and Alvin Combrink and Sabino Francesco Roselli and Martin Fabian},
 year={2026},
 eprint={2602.09605},
 archivePrefix={arXiv},
 primaryClass={eess.SY},
 url={https://arxiv.org/abs/2602.09605}}

  6. Advances in Multi-Agent Path Finding

    A. Combrink

    For the degree of Licentiate, a compilation of previous work in a larger context and their contribution to the field of Multi-Agent Path Finding.

    Chalmers 2025 Licentiate thesis MAPF
    PDF

  7. Optimal Multi-agent Path Finding in Continuous Time

    A. Combrink, S. Roselli, M. Fabian

    Proposes a correction to CCBS, restoring optimality and termination guarantees for the continuous-time MAPF problem.

    AIJ 2025 Under review MAPF
    DOI
    BibTeX 
    @misc{combrink2025optimalmultiagentpathfinding,
 title={Optimal Multi-agent Path Finding in Continuous Time},
 author={Alvin Combrink and Sabino Francesco Roselli and Martin Fabian},
 year={2025},
 eprint={2508.16410},
 archivePrefix={arXiv},
 primaryClass={cs.MA},
 url={https://arxiv.org/abs/2508.16410}}

  8. A Comparative Study of SMT and MILP for the Nurse Rostering Problem

    A. Combrink, S. Do, K. Bengtsson, S. Roselli, M. Fabian

    A comparison of SMT (Z3) and MILP (Gurobi) solvers for personnel scheduling, showing SMT's promise for real-world rostering problems with varied shifts and constraints.

    CODIT 2025 Personnel Scheduling
    DOI
    BibTeX 
    @INPROCEEDINGS{11321384,
 author={Combrink, Alvin and Do, Stephie and Bengtsson, Kristofer and Roselli, Sabino Francesco and Fabian, Martin},
 booktitle={2025 11th International Conference on Control, Decision and Information Technologies (CoDIT)}, 
 title={A Comparative Study of SMT and MILP for the Nurse Rostering Problem}, 
 year={2025},
 volume={1},
 number={},
 pages={2105-2110},
 keywords={Employee welfare;Medical services;Mathematical models;Personnel;Information technology;Standards;Mathematical programming;Formal verification},
 doi={10.1109/CoDIT66093.2025.11321384}}

  9. Prioritized Planning for Continuous-time Lifelong Multi-Agent Pathfinding

    A. Combrink, S. Roselli, M. Fabian

    CPLP: a fast, sub-optimal planner for continuous-time lifelong multi-agent path finding, tested with up to 1000 volumetric agents for practical, real-world applicability.

    CODIT 2025 MAPF
    DOI
    BibTeX 
    @INPROCEEDINGS{11321711,
 author={Combrink, Alvin and Roselli, Sabino Francesco and Fabian, Martin},
 booktitle={2025 11th International Conference on Control, Decision and Information Technologies (CoDIT)}, 
 title={Prioritized Planning for Continuous-time Lifelong Multi-agent Pathfinding}, 
 year={2025},
 volume={1},
 number={},
 pages={1454-1459},
 keywords={Automation;Robustness;Path planning;Planning;Delays;Time factors;Information technology},
 doi={10.1109/CoDIT66093.2025.11321711}}

  10. Online Conflict-Free Scheduling of Fleets of Autonomous Mobile Robots

    F. Popolizio, M. Vinetti, A. Combrink, S. Roselli, M. P. Fanti, M. Fabian

    A heuristic Lifelong MAPF solver that assigns tasks and computes conflict-free plans for hundreds of agents.

    CASE 2024 MAPF
    DOI
    BibTeX 
    @INPROCEEDINGS{10711693,
 author={Popolizio, Francesco and Vinetti, Martina and Combrink, Alvin and Roselli, Sabino Francesco and Pia Fanti, Maria and Fabian, Martin},
 booktitle={2024 IEEE 20th International Conference on Automation Science and Engineering (CASE)}, 
 title={Online Conflict-Free Scheduling of Fleets of Autonomous Mobile Robots}, 
 year={2024},
 volume={},
 number={},
 pages={3063-3068},
 keywords={Schedules;Job shop scheduling;Processor scheduling;Benchmark testing;Throughput;Real-time systems;Path planning;Mobile robots;Optimization;Autonomous robots},
 doi={10.1109/CASE59546.2024.10711693}}

  11. Discrete-event Based Patient Flow Simulation of an Emergency Surgery Department

    A. Combrink, D. Johnson, P. Moldan, M. Fabian

    Discrete-event simulation of hospital patient flow and resource allocation to identify bottlenecks and support more efficient healthcare operations.

    CODIT 2024 Personnel Scheduling
    DOI
    BibTeX 
    @INPROCEEDINGS{10708150,
 author={Combrink, Alvin and Johnson, David and Moldan, Petr and Fabian, Martin},
 booktitle={2024 10th International Conference on Control, Decision and Information Technologies (CoDIT)},
 title={Discrete-Event Based Patient Flow Simulation of an Emergency Surgery Department},
 year={2024},
 pages={1243-1248},
 keywords={Monte Carlo methods;Hospitals;Simulation;Surgery;Medical services;Data models;Resource management;Information technology;Strain;Qualifications;Discrete-Event Modelling;Patient Flow;Emergency Department},
 doi={10.1109/CoDIT62066.2024.10708150}
}

  12. Automatic Shift Scheduling for Healthcare Personnel using Satisfiability Modulo Theory

    A. Combrink, S. Do

    A MSc. thesis on optimising shift assignments for healthcare personnel, using SMT, MILP, and genetic algorithms.

    Chalmers 2021 MSc. thesis Personnel Scheduling
    PDF
    BibTeX 
    @article{combrink2021automatic,
 title={Automatic shift scheduling for healthcare personnel using satisfiability modulo theory},
 author={Combrink, Alvin and Do, Stephie},
 year={2021}}

  13. Formation Control with Collision Avoidance for Spherical Robots

    A. Combrink, D. Karlsson, D. Pettersson, C. Svernlöv, S. Torstensson, J. Warnqvist

    A BSc. thesis on controlling a fleet to follow the leader while avoiding collisions.

    Chalmers 2019 BSc. thesis Motion Planning
    PDF
    BibTeX 
    @article{combrink2019formationsbildning,
 title={Formationsbildning med kollisionsundvikning f{"o}r sf{"a}riska robotar},
 author={Combrink, Alvin and Karlsson, Daniel and Pettersson, Daniel and Svernl{"o}v, Christoffer and Torstensson, Sarah and Warnqvist, Johanna},
 year={2019}}