Collision-Free Path Planning With Local Obstacle Maps
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current path-planning algorithms for autonomous vehicles face challenges in generating collision-free trajectories that are not limited to static grid nodes and do not guarantee optimality, especially in complex environments with varying obstacle densities.
Innovation Solution
The method involves a computing system that uses a sampling-based approach in continuous space to find feasible paths, incorporates a cost function for effective travel distance, and employs a linked list representation of obstacles within a grid cell, allowing the path to deviate from gridlines and iteratively update search positions based on obstacle maps.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If a sampling-based approach in continuous space is used to find feasible paths, then the path can deviate from static grid nodes and achieve better optimality, but the computational complexity and memory usage increase
Solution Approach 1:
The patent divides the continuous search space into discrete grid cells, transforming the complex continuous path planning problem into a manageable discrete grid-based search problem. This segmentation allows the system to maintain optimality guarantees while reducing computational complexity by working with a structured, divided representation of the environment.
Solution Approach 2:
The patent changes the representation parameters from continuous coordinates to discrete grid cell indices, and from complete obstacle maps to localized obstacle information within each cell. This parameter transformation enables efficient computation while preserving the ability to find optimal collision-free paths through the grid structure.
2Reliability
If a complete global obstacle map is used for path planning, then collision-free paths can be guaranteed, but the memory usage and processing time increase
Solution Approach 1:
The global obstacle map is segmented into local obstacle information associated with individual grid cells. Instead of storing and processing a complete global map, the system maintains only the obstacle data relevant to each local cell, dramatically reducing memory requirements while preserving collision-free path guarantees through the grid-based search algorithm.
Solution Approach 2:
The patent implements local quality by storing obstacle information locally within each grid cell rather than maintaining a complete global obstacle map. Each cell contains only the obstacle data pertinent to its local region, enabling memory-efficient path planning that still guarantees collision-free paths through the structured grid search process.
3Productivity
If the path is constrained to follow gridlines, then the computational complexity is reduced, but the path optimality and flexibility decrease
Solution Approach 1:
The patent introduces dynamics by allowing the path to transition from strictly following gridlines to enabling diagonal and curved movements between grid cells. The grid structure provides computational efficiency for search and planning, while the trajectory generation module adds dynamic flexibility to create optimal paths that are not constrained to axial gridline movements.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
According to various aspects, a method of generating a collision free path of travel may include defining a global search area encompassing at least a global start position and a global target position; and determining a set of collision free trajectories by iteration, the set of collision free trajectories connecting the global start position to the global target position via one or more local target positions, each iteration including: determining a local search area within the global search area; determining, from the global obstacle map, a local obstacle map associated with the local search area; defining a local start position and one or more local target positions within the local search area; and calculating, in the local search area, a collision free trajectory from the local start position to the one or more local target positions considering the local obstacle map.