Ranulfo Bezerra

Ranulfo Bezerra·Assistant Professor·Tohoku University

Yamazaki–Bezerra Laboratory

Connected
Robotics

I study how robots, distributed sensors, and edge AI can build shared representations of the physical world, reason together, and coordinate action.

Toward Distributed Physical Intelligence.

ResearchPublicationsAbout
Visit Yamazaki–Bezerra Laboratory

My research within the Yamazaki–Bezerra Laboratory focuses on Connected Robotics, distributed Physical AI, shared world models, and reasoning across robots and sensing systems.

The problem

Devices are connected. Reasoning is not.

Robots, IoT sensors, and edge computers increasingly share data, but their reasoning remains fragmented: each device reaches its own conclusions from its own view of the world. My research investigates how connected physical agents can share evidence, maintain common world models, reason under uncertainty, and generate coordinated actions.

How can connected physical agents think together?

Explore the research

01/Research architecture

  1. 01
    Collective Observability & Adaptive Sensing

    No single agent sees enough. Observability is something a system arranges, not something a sensor has.

    Representative
    MoCArU: Low-Cost Wireless Portable Robot Localization System Using IoT — IEEE SMC 2023, 2023
    Project
    Distributed Localization Infrastructure·Disaster Prevention Challenge, World Robot Summit 2020
  2. 02
    Shared Semantic World Models

    A shared representation is the substrate of collective thought — and it must carry uncertainty and provenance, not just geometry.

    Representative
    Semantic Enrichment of CAD-Based Industrial Environments via Scene Graphs for Simulation and Reasoning — IEEE SSRR 2025, 2025
    Project
    Semantic Scene Graphs for Industrial Environments·Knowledge Acquisition from Sparse Mobile Probe Data
  3. 03
    Distributed Physical Reasoning

    Connected reasoning is the open problem. Sharing data is solved; sharing conclusions is not.

    Representative
    AI-IoT-Robotics Integration: Survey of Frameworks, Emerging Trends, and the Path Toward Connected Robotics — IEEE Internet Things J., 2026
    Project
    Knowledge Acquisition from Sparse Mobile Probe Data
  4. 04
  5. 05
    Edge Physical AI

    Reasoning that cannot run on the robot is not reasoning the robot can use.

    Representative
    AI-IoT-Robotics Integration: Survey of Frameworks, Emerging Trends, and the Path Toward Connected Robotics — IEEE Internet Things J., 2026
    Project
    Semantic Scene Graphs for Industrial Environments

02/Current research

3D scene graph visualisation of a CAD industrial environment, with semantic nodes and relations overlaid on plant structures.

2025 —

Semantic Scene Graphs for Industrial Environments

Turning CAD models into structured, queryable world representations that robots and reasoning systems can share.

02 Shared Semantic World Models · 05 Edge Physical AI

Supervised student: Nathan Pascal Walus

Read more

2023 —

Distributed Localization Infrastructure

Low-cost sensing infrastructure that robots and IoT devices share, so position becomes a property of the environment rather than the robot.

01 Collective Observability & Adaptive Sensing

Read more

Diagram of a transformable garment production floor: mobile robots, a task allocator, static robots, and human workers.

2021 — 2024

Automation for Transformable Production

Multi-robot task allocation and control for garment manufacturing that has to reconfigure itself for every order.

04 Explainable Collective Decision & Action

Read more

Isometric diagram of a road with autonomous vehicles and pedestrian trajectories over stacked knowledge, flow, data, and structure map layers.

2018 — 2021

Knowledge Acquisition from Sparse Mobile Probe Data

Extracting durable environmental knowledge from sparse, incidental observations made by vehicles going about their business.

03 Distributed Physical Reasoning · 02 Shared Semantic World Models

Read more

Industrial plant piping and pumps with a tracked inspection robot in the foreground and three helmeted workers observing.

2020 — 2022

Disaster Prevention Challenge, World Robot Summit 2020

A heterogeneous robot team inspecting an industrial plant — the field test that made the coordination problem concrete.

04 Explainable Collective Decision & Action · 01 Collective Observability & Adaptive Sensing

Read more

All projects

03/Selected publications

62 records·2015–2026

Connected Robotics

Distributed Robot Intelligence

Semantic World Models

Robot-IoT & Distributed Sensing

All publications

04/Research trajectory

How the direction developed. Each stage is backed by publications; earlier work is shown for what it was.

  1. 2015 — 2019

    Semantic Mapping

    Building representations of space from imperfect sensors.

  2. 2018 — 2021

    Knowledge Acquisition

    Extracting durable structure from sparse, incidental observation.

  3. 2021 — 2024

    Multi-Robot Systems

    Coordinating unlike agents toward a shared physical goal.

  4. 2023 —

    Connected Robotics

    Robots, IoT sensors, and edge AI as one reasoning system.

  5. Ongoing

    Distributed Physical Intelligence

    Shared world models and collective reasoning as a scientific programme.

05/Research at Tohoku University

Yamazaki–Bezerra Laboratory

This research is conducted within the Yamazaki–Bezerra Laboratory (Tohoku Robotic Systems Laboratory) at Tohoku University, led by Prof. Kimitoshi Yamazaki, where complementary research spans robotic manipulation, mobile manipulation, household robotics, Connected Robotics, and Physical AI. Connected Robotics and Edge AI is one of the laboratory’s four research pillars, and the direction my own research develops. Part of the work is carried out with the Tough Cyberphysical AI Research Center.

Visit Laboratory Website

06/Recent news