Vehicle Maneuvering Trajectory Selection Using ML Scoring
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing guidance systems for vehicles do not efficiently generate maneuvering trajectories from a vehicle's current location to a goal point on a global path, such as a road or parking space, as they typically consider a location on the global path closest to the vehicle's current location rather than the vehicle's exact location.
Innovation Solution
A method and system utilizing a Machine Learning (ML) model to determine dynamic profile parameters and generate a trajectory score for candidate maneuvering trajectories, selecting the most efficient final trajectory for navigation, which includes an Electronic Control Unit (ECU) configured to obtain candidate trajectories, determine dynamic parameters, and generate a trajectory score to navigate the vehicle to its goal point.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If existing guidance systems generate trajectories from a point on the global path closest to the vehicle rather than from the vehicle's exact current location, then the system complexity is reduced and global path planning is simplified, but the navigation precision and efficiency to reach the goal point deteriorates
Solution Approach 1:
The navigation process is divided into two distinct segments: global path planning (source to goal point) and local maneuvering trajectory generation (vehicle current location to goal point). This segmentation allows each segment to be optimized independently, with the local segment considering the vehicle's exact position while the global segment maintains overall route efficiency.
Solution Approach 2:
The patent applies local quality by generating trajectories specifically tailored to the vehicle's current local position and conditions, rather than using a generic approach from the closest global path point. This enables the system to adapt to local environmental factors, vehicle state, and precise positioning requirements.
2Productivity
If multiple candidate trajectories are generated and evaluated using dynamic profile parameters and ML models, then the navigation efficiency and trajectory selection accuracy are improved, but the computational complexity and processing time increase
Solution Approach 1:
The system performs preliminary actions by pre-generating multiple candidate trajectories and pre-computing their dynamic profile parameters before final selection. This allows the ML model to evaluate pre-processed data, reducing real-time computational burden while maintaining high navigation efficiency.
Solution Approach 2:
The patent introduces an intermediary ML model that acts as a mediator between the candidate trajectories and the final selection process. The ML model evaluates dynamic profile parameters and generates trajectory scores, simplifying the complex decision-making process and enabling efficient trajectory selection based on learned patterns.
3Measurement precision
If the system considers the vehicle's exact current location instead of a generic point on the global path, then the trajectory accuracy and navigation precision are improved, but the computational resources and processing complexity increase
Solution Approach 1:
The patent applies partial action by generating multiple candidate trajectories with varying degrees of optimality rather than computing a single perfect trajectory. This allows the system to find a sufficiently good solution with less computational effort, balancing accuracy requirements with energy constraints.
Data Source
AI summary
The present disclosure relates to generating maneuvering trajectories using Machine Learning (ML) model. The ML model scores each candidate maneuvering trajectories based on a plurality of dynamic profile parameters. A candidate maneuvering trajectory from a plurality of candidate maneuvering trajectories, having a best trajectory score is selected as the final maneuvering trajectory and is provided to the vehicle for navigating according to the final maneuvering trajectory. The final maneuvering trajectory is the most ideal trajectory as it is generated using the ML model based on the plurality of dynamic profile parameters. Also, the vehicle navigation is most efficient when navigated according to the final maneuvering trajectory.