Self-Supervised Learning of Camera-based Drivable Surface Roughness

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F68407700%3A21230%2F21%3A00354062" target="_blank" >RIV/68407700:21230/21:00354062 - isvavai.cz</a>

Výsledek na webu

<a href="https://doi.org/10.1109/IV48863.2021.9575288" target="_blank" >https://doi.org/10.1109/IV48863.2021.9575288</a>

DOI - Digital Object Identifier

<a href="http://dx.doi.org/10.1109/IV48863.2021.9575288" target="_blank" >10.1109/IV48863.2021.9575288</a>

Jazyk výsledku

angličtina

Název v původním jazyce

Self-Supervised Learning of Camera-based Drivable Surface Roughness

Popis výsledku v původním jazyce

A self-supervised method to train a visual predictor of drivable surface roughness in front of a vehicle is proposed. A convolutional neural network taking a single camera image is trained on a dataset labeled automatically by a cross-modal supervision. The dataset is collected by driving a vehicle on various surfaces, while synchronously recording images and accelerometer data. The surface images are labeled by the local roughness measured using the accelerometer signal aligned in time. Our experiments show that the proposed training scheme results in accurate visual predictor. The correlation coefficient between the visually predicted roughness and the true roughness (measured by the accelerometer) is 0.9 on our independent test set of about 1000 images. The proposed method clearly outperforms a baseline method which has the correlation of 0.3 only. The baseline is based on surface texture strength without any training. Moreover, we show a coarse map of local surface roughness, which is implemented by scanning an input image with the trained convolutional network. The proposed method provides automatic and objective road condition assessment, enabling a cheap and reliable alternative to manual data annotation, which is infeasible in a large scale.

Název v anglickém jazyce

Self-Supervised Learning of Camera-based Drivable Surface Roughness

Popis výsledku anglicky

A self-supervised method to train a visual predictor of drivable surface roughness in front of a vehicle is proposed. A convolutional neural network taking a single camera image is trained on a dataset labeled automatically by a cross-modal supervision. The dataset is collected by driving a vehicle on various surfaces, while synchronously recording images and accelerometer data. The surface images are labeled by the local roughness measured using the accelerometer signal aligned in time. Our experiments show that the proposed training scheme results in accurate visual predictor. The correlation coefficient between the visually predicted roughness and the true roughness (measured by the accelerometer) is 0.9 on our independent test set of about 1000 images. The proposed method clearly outperforms a baseline method which has the correlation of 0.3 only. The baseline is based on surface texture strength without any training. Moreover, we show a coarse map of local surface roughness, which is implemented by scanning an input image with the trained convolutional network. The proposed method provides automatic and objective road condition assessment, enabling a cheap and reliable alternative to manual data annotation, which is infeasible in a large scale.

Druh

D - Stať ve sborníku

CEP obor

—

OECD FORD obor

20204 - Robotics and automatic control

Projekt

—

Návaznosti

S - Specificky vyzkum na vysokych skolach

Rok uplatnění

2021

Kód důvěrnosti údajů

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů

Název statě ve sborníku

Proceedings of 2021 IEEE Intelligent Vehicles Symposium (IV)

ISBN

978-1-7281-5394-0

ISSN

—

e-ISSN

—

Počet stran výsledku

7

Strana od-do

1319-1325

Název nakladatele

IEEE

Místo vydání

Piscataway

Místo konání akce

Nagoya

Datum konání akce

11. 7. 2021

Typ akce podle státní příslušnosti

WRD - Celosvětová akce

Kód UT WoS článku

—