Vehicle Path Planning Using Velocity Cones for Dynamic Obstacles
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current path-finding algorithms for autonomous vehicles, such as UAVs and self-driving cars, face challenges in efficiently navigating through environments with static and dynamic obstacles, particularly in real-time applications where quick evasive maneuvers are required, due to high computational demands and inefficiencies in processing data.
Innovation Solution
The proposed solution involves a computer-implemented method and system that generates conflict-free paths by representing obstacles in a 2D+t space, using a velocity cone and polytopes to intersect with obstacles, and computing a visibility graph algorithm to identify conflict-free sub-paths, which reduces computational complexity and allows for efficient navigation through static and dynamic obstacles.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional path-finding algorithms are used to navigate through environments with static and dynamic obstacles, then the vehicle can avoid obstacles, but the computational processing is too slow for real-time applications
Solution Approach 1:
The patent segments the continuous path planning problem into discrete visibility graph nodes and edges. The environment is divided into visible regions from the source, with obstacles represented as polygons that create visibility constraints. This segmentation transforms the continuous navigation problem into a discrete graph traversal problem, enabling faster computational processing while maintaining obstacle avoidance reliability.
Solution Approach 2:
The patent introduces a temporal dimension to the traditional 2D path planning by representing dynamic obstacles as time-varying polygons. The visibility graph is constructed in space-time, where edges represent valid trajectories that avoid obstacles at all times. This dimensional extension allows real-time handling of dynamic obstacles without sacrificing computational efficiency.
2Reliability
If complex algorithms are used to handle moving and morphing obstacles, then the path planning becomes more accurate, but the computational complexity increases
Solution Approach 1:
The patent creates a simplified copy of the dynamic environment as a visibility graph, where complex obstacle motions are represented as static graph structures in space-time. The actual continuous obstacle trajectories are copied into discrete visibility edges that capture the essential avoidance constraints. This copying approach maintains planning accuracy while reducing algorithmic complexity by working with the simplified graph representation rather than continuous obstacle models.
Solution Approach 2:
The patent transforms the problem parameters from continuous obstacle positions and velocities to discrete visibility graph properties (nodes, edges, weights). By changing the representation parameters from physical obstacle states to graph topological features, the algorithm achieves efficient computation while preserving the essential constraints for accurate path planning.
3Productivity
If traditional visibility graph algorithms are used, then paths can be found, but the number of waypoints and sub-paths is excessive
Solution Approach 1:
The patent extracts only the essential visibility relationships from the complete environment model. Instead of generating all possible waypoints and sub-paths, the algorithm identifies and retains only those visibility edges that represent valid, obstacle-free trajectories. This extraction process removes redundant waypoints and sub-paths while preserving the necessary path finding capability, resulting in a more efficient representation.
Data Source
AI summary
A computer-implemented method and a system for generating a path for a vehicle from a source to a target within a two-dimensional (2D) environment with one or more obstacles is disclosed. The obstacles may be dynamic, static or both. The method comprises generating, in a two dimensions-plus-time space, a velocity cone that represents a set of potential waypoints reachable from a first source for the vehicle moving at a speed, providing a polytope, obtaining at least one interception polygon by intersecting and projecting the velocity cone with the polytope on the 2D region; generating a 2D scene comprising interception polygons to avoid, computing a visibility graph algorithm for the 2D scene and obtaining a plurality of conflict-free sub-paths, and composing a valid path connecting the source to the target based on the plurality of conflict-free sub-paths.