Episodic Koopman Learning of Nonlinear Robot Dynamics with Application to Fast Multirotor Landing

2020/04/03 by Folkestad, Carl, Pastor, Daniel, Burdick, Joel W. · 2 citations
#FOS: Electrical engineering #Systems and Control (eess.SY) #electronic engineering #information engineering

paper · doi:10.48550/arxiv.2004.01708

Abstract

This paper presents a novel episodic method to learn a robot's nonlinear dynamics model and an increasingly optimal control sequence for a set of tasks. The method is based on the \em Koopman operator approach to nonlinear dynamical systems analysis, which models the flow of \em observables in a function space, rather than a flow in a state space. Practically, this method estimates a nonlinear diffeomorphism that lifts the dynamics to a higher dimensional space where they are linear. Efficient Model Predictive Control methods can then be applied to the lifted model. This approach allows for real time implementation in on-board hardware, with rigorous incorporation of both input and state constraints during learning. We demonstrate the method in a real-time implementation of fast multirotor landing, where the nonlinear ground effect is learned and used to improve landing speed and quality.

Cited by

• Real-Time Learning of Predictive Dynamic Obstacle Models for Robotic Motion Planning
• Koopman Control Factorization: Data-Driven Convex Controller Design for a Class of Nonlinear Systems

Related

• Learning Koopman Operators with Control Using Bi-level Optimization
• Koopman NMPC: Koopman-based Learning and Nonlinear Model Predictive\n Control of Control-affine Systems
• Koopman Operators in Robot Learning
• Quadrotor Trajectory Tracking with Learned Dynamics: Joint Koopman-based Learning of System Models and Function Dictionaries
• Adaptive Koopman Model Predictive Control of Simple Serial Robots