Cleiton Neves Santos

I study how teams of robots decide, in real time, who intercepts what.

I'm a PhD candidate in Computer Science at the Federal University of Minas Gerais (UFMG), Brazil, working at VeRLab with Prof. Douglas G. Macharet. My thesis treats multi-robot perimeter defense as a dynamic vehicle routing problem, pairing complexity results with algorithms that come with performance guarantees.

I came to robotics from mathematics and from industry software work, so I care about whether an idea holds up both on paper and in running code.

Research and publications

Intruders arrive over time at the boundary of a region, and a team of robots has to decide which defender goes after which intruder, in what order, and when to work alone or join forces.

Team formation and coalitions

Some intruders are more likely to evade capture. I study when defenders should form coalitions and how to keep those coalitions flexible as new threats appear, using probabilistic capture models.

    Adaptive Team Formation for Collaborative Perimeter Defense M.Sc. thesis, UFMG, 2025. 3rd place in the Thesis and Dissertation Contest of the Brazilian Robotics Symposium (SBR 2025); finalist for the Master's Thesis Award at the Brazilian Symposium on Operations Research (SBPO 2026). Video SBPO abstract Award news (in Portuguese)

Homogeneous teams

  • Dynamic Coalition Formation for Probabilistic Perimeter Defense with Homogeneous Defenders C. N. Santos and D. G. Macharet. Brazilian Conference on Robotics (CROS), 2026. DOI

Heterogeneous teams

Teams mix robots with different speeds and capabilities. I give an exact formulation of the allocation problem as an unsplittable flow, prove it is NP-complete and W[1]-hard

  • Manuscript under review C. N. Santos and D. G. Macharet. Available on request

Multiple regions of interest

Often there is not one perimeter to protect but several, spread across the environment. Defenders then have to decide how to split the team across regions as intruders arrive at each of them.

  • Multi-agent Perimeter Defense of Spatially Distributed Regions L. R. D. Paes, C. N. Santos and D. G. Macharet. Brazilian Conference on Robotics (CROS), 2026. DOI

Moving perimeters

When the protected region itself moves, assignments that were feasible a moment ago can stop being so. I am developing a two-layer policy that keeps feasibility guarantees in this setting, evaluated under simulated actuator limits.

  • Manuscript in preparation C. N. Santos and D. G. Macharet.

Experience

  • Full-stack engineer, Multivisão projectVeRLab, UFMG, in partnership with Petrobras. Detecting divergences between 3D point clouds of industrial plants and their CAD models.
    May 2023 – present
  • Developer analystGrupo LPJ IT Consulting, Belo Horizonte. Web platform development with PHP (Laravel), MySQL and AWS.
    Jul 2021 – Mar 2023

Education

  • PhD candidate in Computer ScienceUFMG, advised by Prof. Douglas G. Macharet
    2025 – 2029 (expected)
  • M.Sc. in Computer ScienceUFMG, advised by Prof. Douglas G. Macharet
    2023 – 2025
  • Specialization in Artificial Intelligence and Machine LearningPUC Minas
    2021 – 2022
  • B.Sc. in Computational MathematicsUFMG
    2015 – 2019

Awards

  • Finalist, Master's Thesis AwardBrazilian Symposium on Operations Research (SBPO). Abstract
    2026
  • 3rd place, Thesis and Dissertation Contest, M.Sc. categoryBrazilian Robotics Symposium (SBR), Brazilian Computer Society. DCC/UFMG news
    2025
  • Top 100, Behind the Code Marathon, Latin AmericaIBM
    2021
  • Top 100, Behind the Code Marathon, BrazilIBM
    2019

Talks

  • Adaptive Team Formation for Collaborative Perimeter DefenseMaster's Thesis Award session, Brazilian Symposium on Operations Research (SBPO)
    2026
  • Dynamic Coalition Formation for Probabilistic Perimeter Defense with Homogeneous DefendersBrazilian Conference on Robotics (CROS), João Pessoa
    2026
  • Adaptive Team Formation for Collaborative Perimeter DefenseThesis and Dissertation Contest, Brazilian Robotics Symposium (SBR)
    2025