Dual process system improves autonomous driving safety in unusual situations
Towards Safer Autonomous Driving in an Open World: A Dual-Process Approach
Artificial IntelligenceRobotics
Summary
Self-driving cars usually drive well in normal situations but can struggle when unexpected things happen, like a car running a red light. The researchers behind this paper created a system that uses two ways of planning: one that quickly reacts based on learned experience, and another that carefully thinks through rules and risks for new situations. They also built a part that decides when to switch between these two approaches using knowledge about the environment and traffic rules. Their tests showed this method cuts crashes by a large margin and better follows traffic right-of-way rules.
What this means in practice
- •For autonomous vehicle engineers: Implement a planning system that switches between learned intuition and rule-based reasoning to improve safety in rare driving scenarios.
- •For robotics control teams: Incorporate meta-cognitive risk assessment to manage unfamiliar situations using knowledge graphs and prediction-based control.
Authors
Simon Janssen, Michiel Braat, Chris van der Ploeg, Serge Thill, Jan-Pieter Paardekooper
Abstract
Before autonomous driving systems can be deployed on public roads, it is vital that these systems comply with safety standards, traffic rules, and social norms. Although neural networks trained on large amounts of driving data perform well in routine driving tasks, these models often struggle in novel situations that are not well-represented in the data. In this work, we propose a novel framework that combines a neural network for intuitive, learning-based planning in routine driving tasks with model predictive control for reasoning-based planning in unfamiliar situations, inspired by Dual Process Theory. A meta-cognitive component is designed to switch between the two, using a knowledge graph to reason about contextual risk based on explicit perceptual information and relevant traffic rules and social norms. Contextual risk is represented through risk fields, guiding both the switching mechanism in the meta-cognitive component and compliance with safety standards, traffic rules, and social norms in the reasoning-based planner. The effectiveness of our framework is tested in CARLA for variations of a typical out-of-distribution situations involving (emergency) vehicles running a red light. We show that the novel architecture reduces the number of collisions in the scenarios by 89% and improves compliance with the special right-of-way rules, compared to the NN-only planner.