This thesis examines how a smart edge server, that communicates with autonomous vehicles in an intersection, could contribute to the safety of autonomous driving. Safe path planning is possible when an autonomous vehicle is aware of its surroundings such as obstacles, pedestrians, and other vehicles. Occlusions in intersections caused by, for instance, buildings, trees, and parked vehicles affect the sensing of autonomous vehicles, which, in turn, lead to unsafe autonomous driving. Having a smart edge server that senses and processes useful information in the intersection such as pedestrians, obstacles, and other vehicles, and shares it with autonomous vehicles will be beneficial for safe autonomous driving. However, an important aspect that affects safety is the delay caused by processing the information as well as the network latency. The path planning of an autonomous vehicle will not be accurate nor safe if the information of its surroundings is delayed and old. The objective of this thesis is to implement a latency-aware edge server that communicates with autonomous vehicles and provides them information about their surroundings. For safe autonomous driving, an obstacle avoidance Nonlinear Model Predictive Control, NMPC, algorithm is implemented using the information from the edge server. In order to avoid collisions in an intersection, the edge server uses reachability analysis to compute the set of future possible states for each autonomous vehicle as zonotopes. These reachable sets along with static obstacles in the intersection are broadcasted as useful information. Each autonomous vehicle in the intersection listens to the edge server and adds those obstacles and other vehicles’ reachable sets as unsafe sets in its obstacle avoidance NMPC formulation to plan a safe trajectory. The implementations are evaluated for different scenarios and it is shown that a latency-aware edge server reinforces safety for autonomous driving. The result shows that autonomous vehicles, that have intersecting paths, enter ...


    Zugriff

    Download


    Exportieren, teilen und zitieren



    Titel :

    Latency-aware edge server for safe autonomous driving


    Beteiligte:
    Adnan, Mustafa (Autor:in)

    Erscheinungsdatum :

    2021-01-01


    Medientyp :

    Hochschulschrift


    Format :

    Elektronische Ressource


    Sprache :

    Englisch



    Klassifikation :

    DDC:    629



    Self-Aware Trajectory Prediction for Safe Autonomous Driving

    Shao, Wenbo / Li, Jun / Wang, Hong | IEEE | 2023


    Delay-aware Robust Control for Safe Autonomous Driving

    Kalaria, Dvij / Lin, Qin / Dolan, John M. | IEEE | 2022


    Delay-Aware Robust Control for Safe Autonomous Driving and Racing

    Kalaria, Dvij / Lin, Qin / Dolan, John M. | IEEE | 2024


    UBOL: User-Behavior-aware One-Shot Learning for Safe Autonomous Driving

    Shukla, Sanket / Kasarapu, Sreenitha / Hasan, Rakibul et al. | IEEE | 2022


    Safe and Rule-Aware Deep Reinforcement Learning for Autonomous Driving at Intersections

    Zhang, Chi / Kacem, Kais / Hinz, Gereon et al. | IEEE | 2022