Variable Trust Control Setting for Autonomous Vehicle Highway Navigation and Improved User Experience

Area of Science:

• Robotics and Control Systems
• Human-Computer Interaction
• Transportation Engineering

Background:

• Autonomous vehicles (AVs) require enhanced user trust and acceptance for widespread adoption.
• Ensuring safe and comfortable manoeuvrability, especially during complex interactions, is critical for AV technology.
• Existing control systems often lack mechanisms for user-perceived control, hindering acceptance.

Purpose of the Study:

• To develop and evaluate a model-based design approach for enhancing the acceptance of safe manoeuvrability in autonomous vehicles (AVs) on highways.
• To introduce a variable trust control setting (TCS) that allows users to feel in control, thereby increasing confidence and acceptance.
• To investigate the impact of virtual boundaries (VBs) and steering limitations on AV safety and comfort during overtaking manoeuvres.

Main Methods:

• Development of a variable trust control setting (TCS) grounded in deontological ethics, utilizing virtual boundaries (VBs) for driving decisions.
• Simulation of AVs using a dynamic bicycle model controlled by an adaptive model-predictive control (MPC) algorithm.
• Creation of a novel simulation package to quantify safety metrics for AV interactions and analyze the influence of TCS, VBs, and steering limitations.

Main Results:

• The proposed variable TCS, utilizing VBs, effectively guides AV driving decisions and enhances safety during interactions.
• Simulations demonstrated that VBs and steering limitations significantly influence the safety and comfort of AVs during overtaking.
• The model-based approach proved effective in allowing users to actively manage AV safety and comfort.

Conclusions:

• The variable trust control setting (TCS) is a promising approach to increase user confidence and acceptance of autonomous vehicles.
• Virtual boundaries (VBs) serve as an effective mechanism for ensuring safe and comfortable AV operations, particularly in complex scenarios like overtaking.
• This model-based design approach offers a pathway for developing more user-centric and trustworthy autonomous driving systems.