Trajectory planning for traffic safety with dynamic ethical risk adjustment

Area of Science:

• Autonomous Systems
• Robotics
• AI Ethics

Background:

• Current trajectory planning algorithms for connected and automated vehicles (CAVs) inadequately address ethical risks for all road users and cultural variations.
• Existing methods often prioritize ego-vehicle safety over broader societal values and ethical considerations.

Purpose of the Study:

• To propose a novel trajectory planning algorithm for CAVs that dynamically integrates ethical principles.
• To enable region-specific ethical customization and transparent ethical trade-offs in autonomous systems.

Main Methods:

• Development of an "ethical knob" mechanism for flexible risk weighting between ego vehicles and other road users.
• Implementation of a hybrid subjective-objective weighting method combining Analytic Hierarchy Process and coefficient of variation for ethical principle allocation.
• Integration of ethical frameworks into a safety potential field model to compute risk costs.

Main Results:

• Simulations demonstrated that neutral "ethical knob" settings minimize overall risk costs.
• Dynamic weighting automatically adapts to environmental changes, outperforming static approaches.
• The framework focuses on accident prevention, avoiding predefined ethical dilemmas like the "trolley problem".

Conclusions:

• The proposed algorithm offers culturally adaptable and socially acceptable solutions for autonomous systems.
• This interdisciplinary approach advances the integration of ethical considerations into CAV trajectory planning.
• The framework promotes transparent ethical trade-offs, enhancing trust in autonomous technology.