Gaussian-Process-based Adaptive Tracking Control with Dynamic Active Learning for Autonomous Ground Vehicles

Kristóf Floch, Tamás Péni, and Roland Tóth · 2025

Read Gaussian-Process-based Adaptive Tracking Control with Dynamic Active Learning for Autonomous Ground Vehicles by Kristóf Floch, Tamás Péni, and Roland Tóth (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2501-14672.txt.

Description

Kristóf Floch, Tamás Péni, and Roland Tóth make self-driving control learn from its slips in this two thousand twenty-five paper. Their adaptive trajectory tracking method makes up for modeling errors and unmodeled vehicle dynamics. The base vehicle model splits into lateral and longitudinal subsystems, each topped up with a Gaussian process learned from measurements. Its means feed a compensator that works next to a base feedback controller designed for the system.

An active learning method steers data gathering to telling samples, speeding training. To certify how it does, the team puts forth a counterexample-based method that works out the induced gain between the reference trajectory and tracking error, across a set of vehicle forms. That gives a robust performance certificate under change. The way was shown on a simulator and in real runs with a one-tenth scale F1TENTH race car.

Catalog record

CREATORKristóf Floch, Tamás Péni, and Roland Tóth
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE69 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2501-14672

Browse the full library · About the Tower of Babel