Mohammad Mohades Kasaei
Title: "Data-efficient Non-parametric Modelling and Control of an Extensible Soft Manipulator".
Abstract: Data-driven approaches have shown promising results in modeling and controlling robots, specifically soft and flexible robots where developing physics-based models are more challenging. However, these methods often require a large number of real data, and gathering such data is time-consuming and can damage the robot as well. We proposed a novel data-efficient and non-parametric approach to develop a continuous model using a small dataset of real robot demonstrations (only 25 points). To the best of our knowledge, the proposed approach is the most sample-efficient method for soft continuum robots. Furthermore, we employed this model to develop a controller to track arbitrary trajectories in the feasible kinematic space.
Title: Understanding the roles of insect mushroom body in navigation with an embodied model
Abstract: Despite their tiny brains, many insects are expert navigators. It is well known that they can learn allocentric spatial coordinates of goals (e.g., food) in foraging by a brain structure, called central complex, and later use the memory to take novel shortcuts (e.g., to return nests in a straight line). Some species are also capable of following a familiar route using visual cues. This behaviour cannot be accounted for by the central complex, but it can be explained by another brain structure, called mushroom body, which is responsible for associative learning. In this project, we aim to understand the roles of mushroom body in insect navigation. To validate our mushroom body models, we take an embodied approach by deploying them on a virtual robot in realistic simulations.