Aggressive Deep Driving: Combining Convolutional Neural Networks and Model Predictive Control

Paul Drews; Grady Williams; Brian Goldfain; Evangelos A. Theodorou; James M. Rehg

Aggressive Deep Driving: Combining Convolutional Neural Networks and Model Predictive Control

Paul Drews, Grady Williams, Brian Goldfain, Evangelos A. Theodorou, James M. Rehg

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Abstract

We present a framework for vision-based model predictive control (MPC) for the task of aggressive, high-speed autonomous driving. Our approach uses deep convolutional neural networks to predict cost functions from input video which are directly suitable for online trajectory optimization with MPC. We demonstrate the method in a high speed autonomous driving scenario, where we use a single monocular camera and a deep convolutional neural network to predict a cost map of the track in front of the vehicle. Results are demonstrated on a 1:5 scale autonomous vehicle given the task of high speed, aggressive driving.

Original language	Undefined
Title of host publication	Proceedings of the 1st Annual Conference on Robot Learning
Editors	Sergey Levine, Vincent Vanhoucke, Ken Goldberg
Publisher	PMLR
Pages	133-142
Number of pages	10
Volume	78
State	Published - 2017
Externally published	Yes

Publication series

Name	Proceedings of Machine Learning Research
Publisher	PMLR

Online availability

https://proceedings.mlr.press/v78/drews17a.html

Library availability

Discover UIUC Full Text

Cite this

Drews, P., Williams, G., Goldfain, B., Theodorou, E. A., & Rehg, J. M. (2017). Aggressive Deep Driving: Combining Convolutional Neural Networks and Model Predictive Control. In S. Levine, V. Vanhoucke, & K. Goldberg (Eds.), Proceedings of the 1st Annual Conference on Robot Learning (Vol. 78, pp. 133-142). (Proceedings of Machine Learning Research). PMLR. https://proceedings.mlr.press/v78/drews17a.html

Aggressive Deep Driving: Combining Convolutional Neural Networks and Model Predictive Control. / Drews, Paul; Williams, Grady; Goldfain, Brian et al.
Proceedings of the 1st Annual Conference on Robot Learning. ed. / Sergey Levine; Vincent Vanhoucke; Ken Goldberg. Vol. 78 PMLR, 2017. p. 133-142 (Proceedings of Machine Learning Research).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Drews, P, Williams, G, Goldfain, B, Theodorou, EA & Rehg, JM 2017, Aggressive Deep Driving: Combining Convolutional Neural Networks and Model Predictive Control. in S Levine, V Vanhoucke & K Goldberg (eds), Proceedings of the 1st Annual Conference on Robot Learning. vol. 78, Proceedings of Machine Learning Research, PMLR, pp. 133-142. <https://proceedings.mlr.press/v78/drews17a.html>

@inproceedings{854412cd543e45ae91de9119145ae57c,

title = "Aggressive Deep Driving: Combining Convolutional Neural Networks and Model Predictive Control",

abstract = "We present a framework for vision-based model predictive control (MPC) for the task of aggressive, high-speed autonomous driving. Our approach uses deep convolutional neural networks to predict cost functions from input video which are directly suitable for online trajectory optimization with MPC. We demonstrate the method in a high speed autonomous driving scenario, where we use a single monocular camera and a deep convolutional neural network to predict a cost map of the track in front of the vehicle. Results are demonstrated on a 1:5 scale autonomous vehicle given the task of high speed, aggressive driving.",

author = "Paul Drews and Grady Williams and Brian Goldfain and Theodorou, {Evangelos A.} and Rehg, {James M.}",

year = "2017",

language = "Undefined",

volume = "78",

series = "Proceedings of Machine Learning Research",

publisher = "PMLR",

pages = "133--142",

editor = "Sergey Levine and Vincent Vanhoucke and Ken Goldberg",

booktitle = "Proceedings of the 1st Annual Conference on Robot Learning",

}

TY - GEN

T1 - Aggressive Deep Driving: Combining Convolutional Neural Networks and Model Predictive Control

AU - Drews, Paul

AU - Williams, Grady

AU - Goldfain, Brian

AU - Theodorou, Evangelos A.

AU - Rehg, James M.

PY - 2017

Y1 - 2017

N2 - We present a framework for vision-based model predictive control (MPC) for the task of aggressive, high-speed autonomous driving. Our approach uses deep convolutional neural networks to predict cost functions from input video which are directly suitable for online trajectory optimization with MPC. We demonstrate the method in a high speed autonomous driving scenario, where we use a single monocular camera and a deep convolutional neural network to predict a cost map of the track in front of the vehicle. Results are demonstrated on a 1:5 scale autonomous vehicle given the task of high speed, aggressive driving.

AB - We present a framework for vision-based model predictive control (MPC) for the task of aggressive, high-speed autonomous driving. Our approach uses deep convolutional neural networks to predict cost functions from input video which are directly suitable for online trajectory optimization with MPC. We demonstrate the method in a high speed autonomous driving scenario, where we use a single monocular camera and a deep convolutional neural network to predict a cost map of the track in front of the vehicle. Results are demonstrated on a 1:5 scale autonomous vehicle given the task of high speed, aggressive driving.

M3 - Conference contribution

VL - 78

T3 - Proceedings of Machine Learning Research

SP - 133

EP - 142

BT - Proceedings of the 1st Annual Conference on Robot Learning

A2 - Levine, Sergey

A2 - Vanhoucke, Vincent

A2 - Goldberg, Ken

PB - PMLR

ER -