Autonomous Vehicle Path Library for Real-Time Local Planning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous driving technologies face challenges in navigating tractor-trailer vehicles through complex and restricted geographical spaces, such as narrow streets and parking lots, due to the high computational complexity of path planning and the need for real-time collision avoidance and smooth trajectory management.
Innovation Solution
The method involves generating a compact library of optimal paths offline using a grid node map and orientation grid bin map, performing an exhaustive search for optimal vehicle driving trajectories, and storing these paths on the autonomous vehicle for real-time navigation, reducing the computational burden on onboard resources and enhancing path planning efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If path planning is performed in real-time using onboard computational resources, then the autonomous vehicle can adapt to dynamic environments and avoid collisions, but the computational complexity becomes too high for real-time processing
Solution Approach 1:
The patent pre-computes optimal paths offline using exhaustive search algorithms and stores them in a path library before the vehicle operates. During real-time operation, the vehicle simply retrieves pre-computed paths based on its current state, avoiding the need for complex real-time path planning computations while maintaining collision avoidance capability
Solution Approach 2:
The patent segments the continuous path planning problem into discrete state-space grid nodes and pre-computes optimal transitions between them offline. This segmentation allows the complex continuous optimization problem to be broken down into discrete, pre-solvable subproblems that can be stored and retrieved efficiently during real-time operation
2Manufacturing precision
If exhaustive search is performed for optimal vehicle driving trajectories, then path planning accuracy is improved, but the computational time and resources required increase significantly
Solution Approach 1:
The patent performs the computationally intensive exhaustive search for optimal trajectories in advance during an offline phase, storing the results in a path library. During real-time vehicle operation, the system only needs to retrieve pre-computed optimal paths based on current state, achieving high path planning accuracy without the time penalty of performing exhaustive search in real-time
3Productivity
If a pre-stored path library is used for navigation, then computational load on onboard resources is reduced, but the ability to handle unseen or dynamic environments may be limited
Solution Approach 1:
The patent extends the traditional 2D path planning problem into a 3D state-space that includes position (x, y) and orientation (θ) dimensions. By pre-computing optimal paths in this expanded state-space and storing them in the path library, the system maintains high adaptability to different vehicle states and environmental conditions while keeping real-time computational load low through efficient retrieval operations
Data Source
AI summary
Methods, systems and apparatus for autonomous vehicle path planning and path navigation are described. One example system includes an offline server configured to generate a library of optimal paths for navigating a geographic area, wherein the geographic area is represented as a grid node map and an orientation grid bin map and wherein the optimal paths correspond to paths between pairs of grid node pairs in the grid node map based on optimization criteria, a storage device on the autonomous vehicle for storing the library of optimal paths, and an online server located on the autonomous vehicle configured to access information from the library of optimal paths from the storage device based on a current position and a current heading of the autonomous vehicle, and navigating the autonomous vehicle through the geographic area based on the information.