Human Driving Rule Books for Autonomous Vehicle Decisions
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous vehicles face challenges in making real-time driving decisions, particularly when encountering uncontrolled traffic intersections, stationary vehicles, and jaywalkers, requiring significant processing power or network connectivity, which can be inefficient and increase computational burdens.
Innovation Solution
Generating models of human driving behavior using neural networks and data analysis to create rules for autonomous vehicle maneuvers in various scenarios, such as merging at uncontrolled intersections, passing stationary vehicles, and avoiding jaywalkers, allowing for reduced processing power and network connectivity usage.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If autonomous vehicle systems use sensor data and real-time environmental analysis for decision making, then the safety and reliability of driving decisions is improved, but the computational burden and processing power requirements increase
Solution Approach 1:
The system performs preliminary actions by collecting and analyzing human driving behavior data in advance to generate pre-computed decision rules for various traffic scenarios. These rules are stored in a rule book that can be quickly referenced during real-time operation, eliminating the need for complex real-time computational analysis while maintaining safety and reliability.
Solution Approach 2:
The system creates a simplified copy of human driving decision-making logic by analyzing actual human driver behavior data and encoding it into rule-based representations. This copied knowledge structure allows the autonomous vehicle to make decisions based on pre-established rules rather than performing complex real-time calculations, reducing computational burden while preserving the reliability of human-like decision making.
2Measurement precision
If autonomous vehicle systems perform complex real-time calculations for decision making, then the accuracy of decision making is improved, but the response time and efficiency decrease
Solution Approach 1:
The system performs the computationally intensive analysis of human driving behavior and generation of decision rules in advance, before real-time operation is needed. This preliminary action transfers the computational burden to an offline phase, allowing rapid rule-based decision making during real-time operation without sacrificing accuracy.
Solution Approach 2:
The system replaces the mechanical computation process with a rule-based lookup system. Instead of performing complex real-time calculations, the system substitutes this with pre-computed rules that can be quickly retrieved and applied, maintaining decision accuracy while dramatically reducing response time.
3Reliability
If autonomous vehicle systems rely on network connectivity for decision support, then the quality of decision making is improved, but the system becomes vulnerable to connectivity issues and increases operational complexity
Solution Approach 1:
The system makes itself self-sufficient by encoding human driving behavior knowledge into local rule books that can be executed without external network connectivity. The autonomous vehicle can independently access and apply these pre-computed rules during real-time operation, eliminating vulnerability to connectivity issues while maintaining high-quality decision making.
Solution Approach 2:
The system creates a local copy of human driving knowledge in the form of rule books that are stored in the autonomous vehicle's onboard memory. This copied knowledge structure allows the vehicle to operate independently of network connectivity, removing the vulnerability to connectivity issues while preserving the quality of decision making through access to comprehensive human behavior data.
Data Source
AI summary
The current disclosure provides techniques for using human driving behavior to assist in decision making of an autonomous vehicle as the autonomous vehicle encounters various scenarios on the road. For each scenario, a model may be generated based on human driving behavior that governs how an autonomous vehicle maneuvers in that scenario. As a result of using these models, reliability and safety of autonomous vehicle may be improved. In addition, because the model is programmed into the autonomous vehicle, the autonomous vehicle, in many instances, need not consume resources to implement complex calculations to determine driving behavior in real-time.