Edward Fang

EECS Department, University of California, Berkeley

Technical Report No. UCB/EECS-2020-58

May 25, 2020

http://www2.eecs.berkeley.edu/Pubs/TechRpts/2020/EECS-2020-58.pdf

Autonomous vehicle technology is posed to increase safety and mobility while decreasing emissions and traffic in the transportation industry. Safe, reliable and comfortable autonomous vehicle technology requires the cooperation of robust driving algorithms, failsafe hardware and real-time operating systems. While all components are important, safety-critical motion planning algorithms are paramount as they are responsible for maneuvering the car through complex, dynamic environments.

Motion planning is still an unsolved task in the autonomous driving realm. There are a variety of challenges and shortcomings associated with current state of the art systems, one primary concern being static deadline motion planning. The objective of this paper is to explore the tradeoffs of a variety of planning algorithms in a practical, realistic driving environment. We show how static deadlines are a limitation in current autonomous vehicle motion planners and lay the groundwork for future work using a dynamic deadline system to allow flexible time constraints. We propose that allowing flexible time constraints in motion planning has significant improvements in safety, comfort and reliability over static deadline alternatives.

Advisors: Joseph Gonzalez


BibTeX citation:

@mastersthesis{Fang:EECS-2020-58,
    Author= {Fang, Edward},
    Title= {Dynamic Deadlines in Motion Planning for Autonomous Driving Systems},
    School= {EECS Department, University of California, Berkeley},
    Year= {2020},
    Month= {May},
    Url= {http://www2.eecs.berkeley.edu/Pubs/TechRpts/2020/EECS-2020-58.html},
    Number= {UCB/EECS-2020-58},
    Abstract= {Autonomous vehicle technology is posed to increase safety and mobility while decreasing emissions and traffic in the transportation industry. Safe, reliable and comfortable autonomous vehicle technology requires the cooperation of robust driving algorithms, failsafe hardware and real-time operating systems. While all components are important, safety-critical motion planning algorithms are paramount as they are responsible for maneuvering the car through complex, dynamic environments.

Motion planning is still an unsolved task in the autonomous driving realm. There are a variety of challenges and shortcomings associated with current state of the art systems, one primary concern being static deadline motion planning. The objective of this paper is to explore the tradeoffs of a variety of planning algorithms in a practical, realistic driving environment. We show how static deadlines are a limitation in current autonomous vehicle motion planners and lay the groundwork for future work using a dynamic deadline system to allow flexible time constraints. We propose that allowing flexible time constraints in motion planning has significant improvements in safety, comfort and reliability over static deadline alternatives.},
}

EndNote citation:

%0 Thesis
%A Fang, Edward 
%T Dynamic Deadlines in Motion Planning for Autonomous Driving Systems
%I EECS Department, University of California, Berkeley
%D 2020
%8 May 25
%@ UCB/EECS-2020-58
%U http://www2.eecs.berkeley.edu/Pubs/TechRpts/2020/EECS-2020-58.html
%F Fang:EECS-2020-58