Open access
Date
2022Type
- Conference Paper
ETH Bibliography
yes
Altmetrics
Abstract
Gaussian Process (GP) regressions have proven to be a valuable tool to predict disturbances and model mismatches and incorporate this information into a Model Predictive Control (MPC) prediction. Unfortunately, the computational complexity of inference and learning on classical GPs scales cubically, which is intractable for real-time applications. Thus GPs are commonly trained offline, which is not suited for learning disturbances as their dynamics may vary with time. Recently, state-space formulation of GPs has been introduced, allowing inference and learning with linear computational complexity. This paper presents a framework that enables online learning of disturbance dynamics on quadcopters, which can be executed within milliseconds using a state-space formulation of GPs. The obtained disturbance predictions are combined with MPC leading to a significant performance increase in simulations with jMAVSim. The computational burden is evaluated on a Raspberry Pi 4 B to prove the real-time applicability. Show more
Permanent link
https://doi.org/10.3929/ethz-b-000556758Publication status
publishedExternal links
Book title
2022 American Control Conference (ACC)Pages / Article No.
Publisher
IEEEEvent
Subject
Model Predictive Control (MPC); Quadcopters; Wind disturbance; Machine Learning; Safe learning-based control; Gaussian Process (GP)Organisational unit
02650 - Institut für Automatik / Automatic Control Laboratory03751 - Lygeros, John / Lygeros, John
Notes
Conference lecture held on June 8, 2022
This full text was mistakenly published under a CC BY licence.More
Show all metadata
ETH Bibliography
yes
Altmetrics