Biography
Dr Ioannis Havoutis holds a PhD (2011) and an MSc (2007) from the University of Edinburgh, where he worked on machine learning for motion planning and control of articulated robotic systems.
Ioannis moved to Oxford in March 2017. Previously, he was a postdoctoral researcher at the Robot Learning and Interaction Group of the Idiap Research Institute, Switzerland, where he worked on online learning of complex skills from demonstration. Before this, Ioannis held a senior postdoctoral position in the Dynamic Legged Systems Lab, at the Advanced Robotics Department of the Italian Institute of Technology. There he led the Locomotion Group within the HyQ team, focusing on dynamic motion planning and control for legged locomotion.
Research Interests
Dr Havoutis’ research focuses the combination of machine learning with dynamic whole-body motion planning and control, targeting robots with arms and legs. Building on his previous work, Ioannis is interested both in the locomotion and manipulation aspects of autonomous robots.
He is a co-lead of the Dynamic Robotic Systems (DRS) group -- an integrated research lab within the Oxford Robotics Institute. He lead the research direction of robotic legged locomotion, designing and implementing of algorithms that enable autonomous legged mobility. His focus is on approaches for dynamic whole-body motion planning and control that allow robots with legs to robustly operate in a variety of challenging domains.
His work aims to answer questions such as: which path should the robot choose to reach its goal? how should the body and the individual joints of the robot move? which footholds and potential handholds maximize the robustness of the locomotion behaviour?
Research Groups
Current Projects
EPSRC-UKRI, Robotics and AI in Nuclear (RAIN) research hub
Field inspection of walking robots in sites such as Sellafield and Fukishima with radiation sensors.
EPSRC-UKRI, Off-Shore Robotics for Certification of assets (ORCA) research hub
Mobile robot mapping for inspection of industrial facilities with a variety of platforms.
EU H2020, MEMMO: Memory of Motion
Learning motion representations to enable fast dynamic trajectory optimization and re-planning.
EU H2020, THING: subTerranean Haptic INvestiGator
Haptic sensing, control and estimation for ANYmal.
EPSRC, New Investigator Award
Robust legged locomotion for autonomous mobility in challenging environments.
DPhil Opportunities
I am open to supervising DPhil students with an interest in legged locomotion, motion planning, trajectory optimization, skill representation and learning by demonstration.
Recent Publications
Motion planning in dynamic environments using context-aware human trajectory prediction
Finean MN, Petrović L, Merkt W, Marković I & Havoutis I (2023), Robotics and Autonomous Systems, 104450-104450
BibTeX
@article{motionplanningi-2023/5,
title={Motion planning in dynamic environments using context-aware human trajectory prediction},
author={Finean MN, Petrović L, Merkt W, Marković I & Havoutis I},
journal={Robotics and Autonomous Systems},
number={104450},
pages={104450-104450},
publisher={Elsevier BV},
year = "2023"
}
BiConMP: a nonlinear model predictive control framework for whole body motion planning
Meduri A, Shah P, Viereck J, Khadiv M, Havoutis I et al. (2023), IEEE Transactions on Robotics, 1-18
You Only Look at One: Category-Level Object Representations for Pose Estimation From a Single Example
Goodwin W, Havoutis I & Posner I (2023), Proceedings of Machine Learning Research, 205, 1435-1445
BibTeX
@inproceedings{youonlylookaton-2023/1,
title={You Only Look at One: Category-Level Object Representations for Pose Estimation From a Single Example},
author={Goodwin W, Havoutis I & Posner I},
pages={1435-1445},
year = "2023"
}
Leveraging Scene Embeddings for Gradient-Based Motion Planning in Latent Space
Yamada J, Hung CM, Collins J, Havoutis I & Posner I (2023), Proceedings - IEEE International Conference on Robotics and Automation, 2023-May, 5674-5680
Learning Low-Frequency Motion Control for Robust and Dynamic Robot Locomotion
Gangapurwala S, Campanaro L & Havoutis I (2023), Proceedings - IEEE International Conference on Robotics and Automation, 2023-May, 5085-5091
You Only Look at One: Category-Level Object Representations for Pose Estimation From a Single Example
Goodwin W, Havoutis I & Posner I (2023), Proceedings of Machine Learning Research, 205, 1435-1445
BibTeX
@inproceedings{youonlylookaton-2023/1,
title={You Only Look at One: Category-Level Object Representations for Pose Estimation From a Single Example},
author={Goodwin W, Havoutis I & Posner I},
pages={1435-1445},
year = "2023"
}
Leveraging Scene Embeddings for Gradient-Based Motion Planning in Latent Space
Yamada J, Hung CM, Collins J, Havoutis I & Posner I (2023), Proceedings - IEEE International Conference on Robotics and Automation, 2023-May, 5674-5680
Learning Low-Frequency Motion Control for Robust and Dynamic Robot Locomotion
Gangapurwala S, Campanaro L & Havoutis I (2023), Proceedings - IEEE International Conference on Robotics and Automation, 2023-May, 5085-5091
Roll-Drop: accounting for observation noise with a single parameter
Campanaro L, De Martini D, Gangapurwala S, Merkt W & Havoutis I (2023), Proceedings of Machine Learning Research, 211, 718-730
BibTeX
@inproceedings{rolldropaccount-2023/1,
title={Roll-Drop: accounting for observation noise with a single parameter},
author={Campanaro L, De Martini D, Gangapurwala S, Merkt W & Havoutis I},
pages={718-730},
year = "2023"
}
Asymptotically Optimized Multi-Surface Coverage Path Planning for Loco-Manipulation in Inspection and Monitoring
Ly KT, Munks M, Merkt W & Havoutis I (2023), IEEE International Conference on Automation Science and Engineering, 2023-August