Neural Network Development for IARC Mission 7

This video explains the neural network used for the 7th mission of the International Aerial Robotics Competition. In this competition, the quadrotor must select and ‘bump’ ground vehicles in order to change their velocity and herd them through a gate. Since the trajectories of the robots change on random intervals, the task is complex and difficult to solve. Machine learning was used in order to train an algorithm to maximize the number of herded robots during a fixed time window.