Autonomous Vehicle Risk Evaluation for Unrealized Collision Probability
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional risk scoring systems for autonomous vehicles fail to accurately estimate future collision probabilities as they do not account for the probability that other drivers or autonomous systems will modify their trajectory in response to potential collisions, leading to inaccurate risk assessments.
Innovation Solution
The implementation of a risk evaluation system that uses machine-learning models to calculate unrealized risk metrics by considering the probability of other entities diverting from their trajectories, incorporating factors like kinematic characteristics, reaction time, and environmental conditions, to provide a more accurate assessment of potential collision risks.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional risk scoring systems are used to evaluate collision probabilities, then the evaluation process is simple, but the accuracy of future collision probability estimation deteriorates because they do not account for trajectory modifications by other drivers
Solution Approach 1:
The patent introduces an intermediary component (risk evaluation system with machine learning models) that mediates between raw trajectory data and collision probability assessment. This intermediary accounts for trajectory modifications by other drivers through probabilistic modeling, thereby improving measurement precision without directly increasing system complexity at the vehicle hardware level
Solution Approach 2:
The system changes parameters by incorporating dynamic variables such as reaction time, kinematic characteristics, and environmental conditions into the risk evaluation. These parameter changes enable more accurate collision probability estimation by reflecting real-world driving behavior modifications that conventional static scoring systems cannot capture
2Reliability
If conventional risk scoring systems are used, then the computational resources required are minimal, but the reliability of navigation safety assessment deteriorates due to inaccurate risk evaluations
Solution Approach 1:
The system performs preliminary risk evaluation by pre-calculating collision probabilities for multiple potential trajectories before the vehicle commits to a path. This preliminary action identifies high-risk scenarios in advance, allowing the navigation system to prioritize safe paths without requiring continuous high-energy computational analysis of all possible trajectories
Solution Approach 2:
The patent applies partial action by focusing computational resources on evaluating only the most probable trajectories and those with highest collision risk, rather than exhaustively analyzing all possible paths. This selective evaluation maintains reliability for critical safety assessments while reducing overall computational energy consumption
Data Source
AI summary
The disclosed technology provides solutions for evaluating risk (e.g., collision risk) associated with different vehicle trajectories through an environment. A process of the disclosed technology can include steps for receiving a perception output, wherein the perception output identifies at least one dynamic entity in an environment, determining a projected trajectory for an autonomous vehicle (AV) based on the perception output, and calculating a risk metric for the AV based on the perception output and the projected trajectory for the AV, wherein the risk metric comprises an unrealized risk score that is based on a probability of future collision between the AV and the dynamic entity. Systems and machine-readable media are also provided.