GPU based model-predictive path control for self-driving vehicles. Chajan, E., Schulte-Tigges, J., Reke, M., Ferrein, A., Matheis, D., & Walter, T. In 2021 IEEE Intelligent Vehicles Symposium (IV), pages 1243–1248, July, 2021. Ieeexpl doi abstract bibtex One central challenge for self-driving cars is a proper path-planning. Once a trajectory has been found, the next challenge is to accurately and safely follow the precalculated path. The model-predictive controller (MPC) is a common approach for the lateral control of autonomous vehicles. The MPC uses a vehicle dynamics model to predict the future states of the vehicle for a given prediction horizon. However, in order to achieve real-time path control, the computational load is usually large, which leads to short prediction horizons. To deal with the computational load, the control algorithm can be parallelized on the graphics processing unit (GPU). In contrast to the widely used stochastic methods, in this paper we propose a deterministic approach based on grid search. Our approach focuses on systematically discovering the search area with different levels of granularity. To achieve this, we split the optimization algorithm into multiple iterations. The best sequence of each iteration is then used as an initial solution to the next iteration. The granularity increases, resulting in smooth and predictable steering angle sequences. We present a novel GPU-based algorithm and show its accuracy and realtime abilities with a number of real-world experiments.
@InProceedings{ Chajan-etAl_IV2021_GPU-based-MPC,
author = {Chajan, Eduard and Schulte-Tigges, Joschua and Reke, Michael and Ferrein, Alexander and Matheis, Dominik and Walter, Thomas},
booktitle = {2021 IEEE Intelligent Vehicles Symposium (IV)},
title = {GPU based model-predictive path control for self-driving vehicles},
year = {2021},
month = {July},
pages = {1243--1248},
doi = {10.1109/IV48863.2021.9575619},
url_ieeexpl = {https://ieeexplore.ieee.org/abstract/document/9575619},
keywords = {ADP;Heuristic algorithms;Computational modeling;Graphics
processing units;Stochastic processes;Prediction
algorithms;Trajectory;Vehicle dynamics;autonomous
driving;GPU;model-predictive control;grid
search;path control;ROS2},
abstract = {One central challenge for self-driving cars is a
proper path-planning. Once a trajectory has been
found, the next challenge is to accurately and
safely follow the precalculated path. The
model-predictive controller (MPC) is a common
approach for the lateral control of autonomous
vehicles. The MPC uses a vehicle dynamics model to
predict the future states of the vehicle for a given
prediction horizon. However, in order to achieve
real-time path control, the computational load is
usually large, which leads to short prediction
horizons. To deal with the computational load, the
control algorithm can be parallelized on the
graphics processing unit (GPU). In contrast to the
widely used stochastic methods, in this paper we
propose a deterministic approach based on grid
search. Our approach focuses on systematically
discovering the search area with different levels of
granularity. To achieve this, we split the
optimization algorithm into multiple iterations. The
best sequence of each iteration is then used as an
initial solution to the next iteration. The
granularity increases, resulting in smooth and
predictable steering angle sequences. We present a
novel GPU-based algorithm and show its accuracy and
realtime abilities with a number of real-world
experiments.},
}
Downloads: 0
{"_id":"whQJHtDFfuaJvfw6B","bibbaseid":"chajan-schultetigges-reke-ferrein-matheis-walter-gpubasedmodelpredictivepathcontrolforselfdrivingvehicles-2021","author_short":["Chajan, E.","Schulte-Tigges, J.","Reke, M.","Ferrein, A.","Matheis, D.","Walter, T."],"bibdata":{"bibtype":"inproceedings","type":"inproceedings","author":[{"propositions":[],"lastnames":["Chajan"],"firstnames":["Eduard"],"suffixes":[]},{"propositions":[],"lastnames":["Schulte-Tigges"],"firstnames":["Joschua"],"suffixes":[]},{"propositions":[],"lastnames":["Reke"],"firstnames":["Michael"],"suffixes":[]},{"propositions":[],"lastnames":["Ferrein"],"firstnames":["Alexander"],"suffixes":[]},{"propositions":[],"lastnames":["Matheis"],"firstnames":["Dominik"],"suffixes":[]},{"propositions":[],"lastnames":["Walter"],"firstnames":["Thomas"],"suffixes":[]}],"booktitle":"2021 IEEE Intelligent Vehicles Symposium (IV)","title":"GPU based model-predictive path control for self-driving vehicles","year":"2021","month":"July","pages":"1243–1248","doi":"10.1109/IV48863.2021.9575619","url_ieeexpl":"https://ieeexplore.ieee.org/abstract/document/9575619","keywords":"ADP;Heuristic algorithms;Computational modeling;Graphics processing units;Stochastic processes;Prediction algorithms;Trajectory;Vehicle dynamics;autonomous driving;GPU;model-predictive control;grid search;path control;ROS2","abstract":"One central challenge for self-driving cars is a proper path-planning. Once a trajectory has been found, the next challenge is to accurately and safely follow the precalculated path. The model-predictive controller (MPC) is a common approach for the lateral control of autonomous vehicles. The MPC uses a vehicle dynamics model to predict the future states of the vehicle for a given prediction horizon. However, in order to achieve real-time path control, the computational load is usually large, which leads to short prediction horizons. To deal with the computational load, the control algorithm can be parallelized on the graphics processing unit (GPU). In contrast to the widely used stochastic methods, in this paper we propose a deterministic approach based on grid search. Our approach focuses on systematically discovering the search area with different levels of granularity. To achieve this, we split the optimization algorithm into multiple iterations. The best sequence of each iteration is then used as an initial solution to the next iteration. The granularity increases, resulting in smooth and predictable steering angle sequences. We present a novel GPU-based algorithm and show its accuracy and realtime abilities with a number of real-world experiments.","bibtex":"@InProceedings{ Chajan-etAl_IV2021_GPU-based-MPC,\n author = {Chajan, Eduard and Schulte-Tigges, Joschua and Reke, Michael and Ferrein, Alexander and Matheis, Dominik and Walter, Thomas},\n booktitle = {2021 IEEE Intelligent Vehicles Symposium (IV)}, \n title = {GPU based model-predictive path control for self-driving vehicles}, \n year = {2021},\n month = {July},\n pages = {1243--1248},\n doi = {10.1109/IV48863.2021.9575619},\n url_ieeexpl = {https://ieeexplore.ieee.org/abstract/document/9575619},\n keywords = {ADP;Heuristic algorithms;Computational modeling;Graphics\n processing units;Stochastic processes;Prediction\n algorithms;Trajectory;Vehicle dynamics;autonomous\n driving;GPU;model-predictive control;grid\n search;path control;ROS2},\n abstract = {One central challenge for self-driving cars is a\n proper path-planning. Once a trajectory has been\n found, the next challenge is to accurately and\n safely follow the precalculated path. The\n model-predictive controller (MPC) is a common\n approach for the lateral control of autonomous\n vehicles. The MPC uses a vehicle dynamics model to\n predict the future states of the vehicle for a given\n prediction horizon. However, in order to achieve\n real-time path control, the computational load is\n usually large, which leads to short prediction\n horizons. To deal with the computational load, the\n control algorithm can be parallelized on the\n graphics processing unit (GPU). In contrast to the\n widely used stochastic methods, in this paper we\n propose a deterministic approach based on grid\n search. Our approach focuses on systematically\n discovering the search area with different levels of\n granularity. To achieve this, we split the\n optimization algorithm into multiple iterations. The\n best sequence of each iteration is then used as an\n initial solution to the next iteration. The\n granularity increases, resulting in smooth and\n predictable steering angle sequences. We present a\n novel GPU-based algorithm and show its accuracy and\n realtime abilities with a number of real-world\n experiments.},\n}\n","author_short":["Chajan, E.","Schulte-Tigges, J.","Reke, M.","Ferrein, A.","Matheis, D.","Walter, T."],"key":"Chajan-etAl_IV2021_GPU-based-MPC","id":"Chajan-etAl_IV2021_GPU-based-MPC","bibbaseid":"chajan-schultetigges-reke-ferrein-matheis-walter-gpubasedmodelpredictivepathcontrolforselfdrivingvehicles-2021","role":"author","urls":{" ieeexpl":"https://ieeexplore.ieee.org/abstract/document/9575619"},"keyword":["ADP;Heuristic algorithms;Computational modeling;Graphics processing units;Stochastic processes;Prediction algorithms;Trajectory;Vehicle dynamics;autonomous driving;GPU;model-predictive control;grid search;path control;ROS2"],"metadata":{"authorlinks":{}}},"bibtype":"inproceedings","biburl":"http://maskor.fh-aachen.de/biblio/MASKOR.bib","dataSources":["i8ftsMK5wMbiNqmtB","pBDNm3knLemYTNMHw"],"keywords":["adp;heuristic algorithms;computational modeling;graphics processing units;stochastic processes;prediction algorithms;trajectory;vehicle dynamics;autonomous driving;gpu;model-predictive control;grid search;path control;ros2"],"search_terms":["gpu","based","model","predictive","path","control","self","driving","vehicles","chajan","schulte-tigges","reke","ferrein","matheis","walter"],"title":"GPU based model-predictive path control for self-driving vehicles","year":2021}