Robots that learn to become adaptive, versatile, and autonomous.
The objective of our lab is to improve the algorithmic foundations of learning algorithms to increase the versatility, resilience and autonomy of physical robots. Our motivation is that robots have the potential to deliver tremendous benefits to society, but they currently only operate in well controlled environment, like factories or warehouses. One of the main reasons for that is that engineers and developers cannot anticipate all the situations that the robots will face when operating in an uncontrolled environment. In the Adaptive & Intelligent Robotics Lab, we develop learning algorithms that enable physical robots to autonomously adapt to new and unforeseen situations, like a new environment, a new task or even a mechanical damage. Our research encompass several topic areas of machine learning and robotics, like reinforcement learning, evolutionary algorithms, deep neural networks, bayesian optimisation, unsupervised learning and optimal control.