The research focus is on developing state-of-the-art machine learning algorithms (including Deep learning, Reinforcement learning, and Unsupervised learning) for intelligent robot behaviour. We aim to advance the decisional autonomy of robots, improve their perception, their understanding of the world, and their prediction capabilities. The ultimate goal is to create algorithms that allow robots to autonomously learn from their environment in an open-ended and continuous way (i.e. lifelong learning).

Robot learns the skill of archery

After being instructed how to hold the bow and release the arrow, the robot learns by itself to aim and shoot arrows at the target. The learning algorithm is called ARCHER (Augmented Reward Chained Regression).

Learning Symbolic Representations of Actions from Human Demonstrations

Imitation learning enables a robot to acquire new trajectory-based skills from demonstrations. This novel machine learning approach integrates imitation learning, Visuospatial Skill Learning, and a symbolic planner.

Interactive robot learning of visuospatial skills

Interactive robot learning of visuospatial skills. The so-called "visuospatial skills" allow people to visually perceive objects and the spatial relationships among them.

Thruster Failure Recovery on Autonomous Underwater Vehicle

Thruster Failure Recovery on Autonomous Underwater Vehicle. The learning approach is able to discover new control policies to overcome thruster failures as they happen, using a model-based direct policy search algorithm.