ML & Articulated Robots Group

We develop learning, planning and control methods that allow articulated robots with arms and legs to move dynamically, manipulate, inspect and operate in challenging real-world environments.

News

ARG represents ORI and Oxford at Oxplore in Birmingham

Mark Bromley and Kai Lu with the Unitree G1 robot at the ORI and Oxford Oxplore stand in Birmingham

This week Mark Bromley and Kai Lu represented ARG, the Oxford Robotics Institute and the University of Oxford at Oxplore in Birmingham, with support from Daniel Marques. They showcased our Unitree G1 humanoid robot and shared the group's work on machine learning and articulated robots.

Read more

MetaSym published in TMLR

Pranav Vaidhyanathan, Aristotelis Papatheodorou, Mark T. Mitchison, Natalia Ares and Ioannis Havoutis publish MetaSym, a symplectic meta-learning framework for physical intelligence.

Read more

New IROS work on adaptive manipulation

ANYmal quadruped robot with a manipulator arm in the Oxford Robotics Institute lab

Jacques Cloete, Wolfgang Merkt and Ioannis Havoutis present adaptive manipulation using behavior trees at IROS.

Read more

Mobile Robotic Inspector project begins

ANYmal quadruped robot with manipulator arm operating a valve for the Mobile Robotic Inspector project

The group starts an EPSRC International Centre-to-Centre collaboration with ETH Zurich and RobotX on loco-manipulation for mobile inspection.

Read more

Robots that move, adapt and work outside the lab.

ARG combines robot learning, optimal control, trajectory optimisation and whole-body planning to build robust behaviours for legged robots and mobile manipulators.

Robot learning

Compact, reusable and adaptive task representations using machine learning and optimisation.

Legged locomotion

Dynamic whole-body motion planning, contact-rich locomotion and robust control for quadrupeds and humanoids.

Loco-manipulation

Integrated mobility and manipulation for inspection, maintenance and agile assembly tasks.