國立虎尾科技大學 |

Human Behavior Modeling and Human Behavior-Aware Control of Automated Vehicles for Trustworthy Navigation.

紀錄類型:	書目-語言資料,印刷品 : Monograph/item
正題名/作者:	Human Behavior Modeling and Human Behavior-Aware Control of Automated Vehicles for Trustworthy Navigation./
作者:	Jayaraman, Suresh Kumaar.
出版者:	Ann Arbor : ProQuest Dissertations & Theses, : 2021,
面頁冊數:	172 p.
附註:	Source: Dissertations Abstracts International, Volume: 83-05, Section: B.
Contained By:	Dissertations Abstracts International83-05B.
標題:	Artificial intelligence. -
電子資源:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=28846635
ISBN:	9798471104235

Human Behavior Modeling and Human Behavior-Aware Control of Automated Vehicles for Trustworthy Navigation.
Jayaraman, Suresh Kumaar.

Human Behavior Modeling and Human Behavior-Aware Control of Automated Vehicles for Trustworthy Navigation. - Ann Arbor : ProQuest Dissertations & Theses, 2021 - 172 p.

Source: Dissertations Abstracts International, Volume: 83-05, Section: B.

Thesis (Ph.D.)--University of Michigan, 2021.

This item must not be sold to any third party vendors.

Robots are increasingly being developed for and used in human-centered environments, and there is a growing need to develop methods for safe human-robot interaction. Automated vehicles (AVs), a type of robot that has garnered much attention recently, can increase transport safety, efficiency, and accessibility. However, to realize these benefits, they need to be trusted and accepted by the general public.This dissertation focuses on the interaction between AVs and pedestrians, one of the most vulnerable types of road users. AVs are a novel technology, and thus the interaction dynamics between AVs and pedestrians are not clear. A major operational challenge for AVs is safe navigation in urban environments around pedestrians. To improve pedestrian trust in AVs, researchers typically develop motion planning methods that can guarantee safe operation. However, in addition to safety, other factors (environmental and behavioral) can influence trust in the AVs and, in turn, their acceptance.Typically, AVs employ a receding-horizon planning methodology, where they plan for a short horizon (1 − 2 s) while incorporating predictions of pedestrian trajectories to avoid potential collisions. The urban traffic environment is dynamic in nature and constantly changes with the location and the behaviors of the surrounding road agents. This requires the AV to plan in real time. The prediction and planning models, therefore, should be computationally efficient to enable real-time planning. A key challenge in AV planning is balancing safety and performance. Focusing only on safety can lead to highly conservative AV behaviors that are undesirable. Further, to gain public trust and acceptance, the AVs should demonstrate navigation capabilities that are both safe and trustworthy. Extending the prediction and planning horizons to a longer term (> 5 s) could aid the AV in developing such safe and trustworthy trajectories.This dissertation addresses two high-level research problems in the context of pedestrian-AV interaction—(i) how to predict long-term pedestrian behaviors efficiently and (ii) how to use the pedestrian behavior predictions to plan safe and trustworthy trajectories in real time. This dissertation has four primary contributions. First, this dissertation characterizes the effects of AV driving behavior and environmental factors on pedestrian’s trust in the AVs and pedestrian behavior, based on user studies developed in virtual and controlled real-world environments. Second, a new modeling framework for urban pedestrian behavior based on hybrid systems theory is presented. The framework models the high-level intent and decision-making process of pedestrians and uses a simple continuous motion model. Third, the framework is extended to include interaction between other pedestrians and predict multimodal pedestrian behaviors. The proposed framework is tested on publicly available realworld datasets and a virtual reality dataset collected from a user study. The results show the model’s ability to predict long-term multimodal pedestrian behaviors that are intuitive and explainable. Finally, a receding-horizon planner that incorporates the pedestrian predictions is presented. The planner was tested in a simulated traffic environment. Results indicate the potential of the approach to developing safe AV behaviors that are understandable and trustworthy. The models and methods discussed in this dissertation enable a better understanding of human and robot behaviors, thereby aiding in realizing safe and trustworthy human-robot interactions.

ISBN: 9798471104235Subjects--Topical Terms:

559380
Artificial intelligence.
Subjects--Index Terms:

Automated vehicles

Human Behavior Modeling and Human Behavior-Aware Control of Automated Vehicles for Trustworthy Navigation.
LDR:04885nam a2200397 4500 001 1067277
005 20220823142335.5
008 221020s2021 ||||||||||||||||| ||eng d
020 $a 9798471104235
035 $a (MiAaPQ)AAI28846635
035 $a (MiAaPQ)umichrackham003795
035 $a AAI28846635
040 $a MiAaPQ $c MiAaPQ
100 1 $a Jayaraman, Suresh Kumaar. $3 1372807
245 1 0 $a Human Behavior Modeling and Human Behavior-Aware Control of Automated Vehicles for Trustworthy Navigation.
260 1 $a Ann Arbor : $b ProQuest Dissertations & Theses, $c 2021
300 $a 172 p.
500 $a Source: Dissertations Abstracts International, Volume: 83-05, Section: B.
500 $a Advisor: Tilbury, Dawn.
502 $a Thesis (Ph.D.)--University of Michigan, 2021.
506 $a This item must not be sold to any third party vendors.
506 $a This item must not be added to any third party search indexes.
520 $a Robots are increasingly being developed for and used in human-centered environments, and there is a growing need to develop methods for safe human-robot interaction. Automated vehicles (AVs), a type of robot that has garnered much attention recently, can increase transport safety, efficiency, and accessibility. However, to realize these benefits, they need to be trusted and accepted by the general public.This dissertation focuses on the interaction between AVs and pedestrians, one of the most vulnerable types of road users. AVs are a novel technology, and thus the interaction dynamics between AVs and pedestrians are not clear. A major operational challenge for AVs is safe navigation in urban environments around pedestrians. To improve pedestrian trust in AVs, researchers typically develop motion planning methods that can guarantee safe operation. However, in addition to safety, other factors (environmental and behavioral) can influence trust in the AVs and, in turn, their acceptance.Typically, AVs employ a receding-horizon planning methodology, where they plan for a short horizon (1 − 2 s) while incorporating predictions of pedestrian trajectories to avoid potential collisions. The urban traffic environment is dynamic in nature and constantly changes with the location and the behaviors of the surrounding road agents. This requires the AV to plan in real time. The prediction and planning models, therefore, should be computationally efficient to enable real-time planning. A key challenge in AV planning is balancing safety and performance. Focusing only on safety can lead to highly conservative AV behaviors that are undesirable. Further, to gain public trust and acceptance, the AVs should demonstrate navigation capabilities that are both safe and trustworthy. Extending the prediction and planning horizons to a longer term (> 5 s) could aid the AV in developing such safe and trustworthy trajectories.This dissertation addresses two high-level research problems in the context of pedestrian-AV interaction—(i) how to predict long-term pedestrian behaviors efficiently and (ii) how to use the pedestrian behavior predictions to plan safe and trustworthy trajectories in real time. This dissertation has four primary contributions. First, this dissertation characterizes the effects of AV driving behavior and environmental factors on pedestrian’s trust in the AVs and pedestrian behavior, based on user studies developed in virtual and controlled real-world environments. Second, a new modeling framework for urban pedestrian behavior based on hybrid systems theory is presented. The framework models the high-level intent and decision-making process of pedestrians and uses a simple continuous motion model. Third, the framework is extended to include interaction between other pedestrians and predict multimodal pedestrian behaviors. The proposed framework is tested on publicly available realworld datasets and a virtual reality dataset collected from a user study. The results show the model’s ability to predict long-term multimodal pedestrian behaviors that are intuitive and explainable. Finally, a receding-horizon planner that incorporates the pedestrian predictions is presented. The planner was tested in a simulated traffic environment. Results indicate the potential of the approach to developing safe AV behaviors that are understandable and trustworthy. The models and methods discussed in this dissertation enable a better understanding of human and robot behaviors, thereby aiding in realizing safe and trustworthy human-robot interactions.
590 $a School code: 0127.
650 4 $a Artificial intelligence. $3 559380
650 4 $a Behavioral sciences. $3 1148596
650 4 $a Automotive engineering. $3 1104081
650 4 $a Robotics. $3 561941
650 4 $a Mechanical engineering. $3 557493
653 $a Automated vehicles
653 $a Behavior model
653 $a Decision-making
690 $a 0771
690 $a 0548
690 $a 0800
690 $a 0602
690 $a 0540
710 2 $a University of Michigan. $b Mechanical Engineering. $3 1178961
773 0 $t Dissertations Abstracts International $g 83-05B.
790 $a 0127
791 $a Ph.D.
792 $a 2021
793 $a English
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=28846635