Research
The ML & Articulated Robots Group studies dynamic motion generation for robots with complex articulated bodies. Our work connects model-based optimisation, reinforcement learning, manifold and probabilistic representations, and online control.
Dynamic motion planning and control
Whole-body approaches for legged robots, optimal control, trajectory optimisation, dynamical systems and robust execution on hardware.
Learning for articulated robots
Manifold learning, geometric representations, compact reusable skills, uncertainty-aware policies and online adaptation.
Loco-manipulation and inspection
Mobile robotic inspection, task-and-motion planning, quadrupedal robots with arms, coverage planning and operation in industrial environments.
Human and robot skill models
Learning reusable motion memories and sensorimotor repertoires for safer robotic behaviour and physical intelligence.
Oxford Indoor Human Motion Dataset
A human trajectory prediction dataset using static and robot-mounted RGB-D cameras to observe humans as they move in an indoor environment.