Autonomous Vehicle Control Evaluation Using Manual Driving Trajectories
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous vehicles face challenges in accurately evaluating their control systems to ensure they operate consistently with human-driven vehicles, particularly in complex environments, due to differences in trajectory planning and reaction times.
Innovation Solution
The use of manual driving data to evaluate autonomous vehicle control systems by comparing predicted trajectories with actual human-driven trajectories, identifying deviations, and adjusting cost functions to improve the accuracy of autonomous vehicle control systems.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If autonomous vehicle control systems are evaluated using extensive real-world testing and simulation, then the reliability of the control system improves, but the loss of time and resources increases
Solution Approach 1:
The patent uses manual driving data as a copy of real-world driving behavior to evaluate the autonomous vehicle control system. Instead of performing extensive real-world testing, the system processes recorded manual driving data to generate evaluation metrics, significantly reducing testing time while maintaining evaluation reliability
Solution Approach 2:
The patent performs preliminary processing of manual driving data to create a standardized evaluation dataset before conducting the actual control system evaluation. This preliminary action includes data cleaning, feature extraction, and trajectory reconstruction, which enables more efficient and focused evaluation processes
2Adaptability or versatility
If the autonomous vehicle control system predicts trajectories that differ significantly from manual driving trajectories, then the adaptability of the system to complex environments improves, but the manufacturing precision of the trajectory planning decreases
Solution Approach 1:
The patent adjusts cost function parameters in the trajectory planning system based on deviations observed between predicted and manual driving trajectories. By dynamically modifying parameters such as safety margins, comfort weights, and trajectory smoothness factors, the system achieves both environmental adaptability and trajectory precision
Solution Approach 2:
The patent implements a feedback mechanism where evaluation results from comparing predicted trajectories with manual driving trajectories are used to iteratively improve the control system. The system continuously refines its cost functions and planning algorithms based on performance metrics derived from this comparison
Data Source
AI summary
Techniques are disclosed for evaluating an autonomous vehicle (“AV”) control system by determining deviations between data generated using the AV control system and manual driving data. In many implementations, manual driving data captures action(s) of a vehicle controlled by a manual driver. Additionally or alternatively, multiple AV control systems can be evaluated by comparing deviations for each AV control system, where the deviations are determined using the same set of manual driving data.