Automated Control Explanations for Transparent Driving Decisions
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Users of automated driving systems often experience uncertainty and concern due to the lack of transparency in decision-making processes, leading to reduced confidence in semi- and fully autonomous vehicles as they may not understand why certain behaviors or maneuvers are performed.
Innovation Solution
A system and method that generate and present explanations to users using a gray box decision-making algorithm, incorporating sensor data, user models, and probabilistic algorithms to provide explanations based on operating contexts, which can be adjusted in length and timing to mitigate user concern, utilizing a Markov Decision Process (MDP) for generating explanations such as myopic, series, contrastive, semantic, or user-centered explanations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If automated control systems use decision-making algorithms to perform driving maneuvers, then the system's automation capability and driving performance are improved, but the transparency of the decision-making process deteriorates, causing users to be unsure about the reasons for certain behaviors
Solution Approach 1:
The patent introduces an explanation module as an intermediary between the automated control system and the user. This module translates the internal decision-making process into human-understandable explanations, allowing users to comprehend the reasons behind automated behaviors without compromising the automation's operational integrity or complexity
2Reliability
If the system provides detailed explanations of its decision-making process, then user confidence and understanding are improved, but the system complexity and computational resources required increase
Solution Approach 1:
The explanation module provides differentiated explanations tailored to specific user needs and contexts rather than uniform detailed explanations for all situations. The system adapts the level and type of explanation based on the particular driving scenario and user preferences, delivering sufficient transparency without uniformly increasing system complexity across all operating conditions
3Reliability
If the system proactively provides explanations based on user models, then user concern is reduced and trust is improved, but the computational processing and time required for explanation generation increase
Solution Approach 1:
The system pre-generates and stores explanations for common driving scenarios and decision situations. When a driving maneuver occurs, the explanation module quickly retrieves the pre-prepared explanation rather than generating it in real-time, significantly reducing the time required to provide explanations while maintaining proactive communication with the user
Data Source
AI summary
A system for generating and presenting an explanation to a user includes an input module configured to receive sensor data and input the sensor data to a control system of an automated system, the control system configured to control operation of the automated system based on the sensor data and according to a gray box decision making algorithm. The system also includes an explanation module configured to generate an explanation of a decision made by the control system and a corresponding behavior of the automated system, the explanation generated based on at least one of: a request by the user, and a user model, and a display module configured to present the generated explanation to the user.