A simple approach for mapping controllers between quadrotors for similar performance

Abstract

Based on the differential flatness property of quadrotors, we propose a simple approach for transferring control policies between different quadrotors. We show that to guarantee identical performance between different quadrotors, a simple non-linear transformation between the control signals of the two quadrotors is sufficient. This transformation depends on the weight and inertia matrix of both quadrotors. We also study the problem of efficiently generating dynamically feasible trajectories for quadrotors. A supervised learning approach is used to train a deep neural network with two hidden layers. The training data is generated from a well-known trajectory generation method that minimizes jerk given a fixed interval time. More than a million dynamically feasible trajectories between two random points in 3D space are generated and used as training data. The input of the neural network is a vector composed of initial and desired states, along with the final trajectory time. The output of the neural network generates the motion primitives of the trajectories, as well as the duration or final time of a segment. Simulation results show very fast dynamically feasible trajectory generation by the proposed deep learning algorithm.