planning-and-decision-making

RO47005 Planning & Decision Making. Quadrotor model planner using probabilistic roadmap (PRM) and collision avoidance using Velocity Obstacles

imitation-learning

Reproduced human hand movements on a simulated manipulator with an R2 score over 89% by imitation learning. Action demonstrations are acquired using a Kinect motion sensor to record hand movements. Firstly, task trajectories are encoded by Gaussian Mixture Model fit by Expectation Maximization. Secondly, motion is retrieved by statistical inference

multi-agent-systems

A set of autonomous agents are coordinated to collaborate on a variety of problems, in a simulated UNITY environment.