Bart Dhoedt

About

Bart Dhoedt is a prominent researcher whose work sits at the intersection of autonomous robotics, machine learning, and biologically inspired artificial intelligence. He is perhaps best known for his pioneering contributions to **active inference** in robotics — a framework rooted in neuroscience that treats perception and action as unified processes of minimizing surprise. His highly cited work on hierarchical active inference for robot navigation (79 citations) and generative state space models (39 citations) has helped establish active inference as a credible alternative to conventional reinforcement learning for autonomous agents. Beyond theoretical contributions, Dhoedt has made substantial practical advances in robotic manipulation, developing data-efficient learning-from-demonstration approaches that enable robots to grasp objects from remarkably few examples. His earlier research in deep reinforcement learning for sensor fusion (26 citations) and distributed middleware for cyber-physical systems laid important groundwork for intelligent, connected robotic platforms. More recently, his group has explored computational optimization of image-based learning and biologically inspired SLAM systems, reflecting a sustained commitment to making sophisticated AI practical on real robotic hardware. Across his body of work, Dhoedt consistently bridges theoretical rigor with real-world applicability, making him a significant voice in the evolving field of intelligent autonomous systems.

Research Focus

Key Achievements

Top Papers

1. 1
   Robot navigation as hierarchical active inference

   79 citations · 2021
2. 2
   Learning Generative State Space Models for Active Inference

   39 citations · 2020
3. 3
   Learning robots to grasp by demonstration

   33 citations · 2020
4. 4
   Sensor fusion for robot control through deep reinforcement learning

   26 citations · 2017
5. 5
   Active Vision for Robot Manipulators Using the Free Energy Principle

   23 citations · 2021
6. 6
   Middleware Platform for Distributed Applications Incorporating Robots, Sensors and the Cloud

   14 citations · 2016
7. 7
   Learning to Grasp from a Single Demonstration

   10 citations · 2018
8. 8
   Deep Active Inference for Autonomous Robot Navigation

   7 citations · 2020
9. 9
   Computational Optimization of Image-Based Reinforcement Learning for Robotics

   6 citations · 2022
10. 10
    LatentSLAM: unsupervised multi-sensor representation learning for localization and mapping

    6 citations · 2021

Key Collaborators

Contact & Links