Autonomous Valet Parking Path Planner Cost Function
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing autonomous parking systems require human intervention to locate and position the vehicle near a parking spot, lacking the ability to autonomously utilize preexisting infrastructure for optimized path planning and mimicking human driving behavior.
Innovation Solution
A method for autonomously parking or un-parking a vehicle by generating a path with linked nodes, where each node's cost is calculated based on inherited and change costs, including turning and direction costs, to optimize path planning and avoid obstacles, using sensors and vehicle control systems for navigation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional path planning algorithms are used for autonomous parking, then the system can navigate to parking spots, but the path planning is inefficient with excessive turning and gear changes
Solution Approach 1:
The patent applies parameter changes by modifying the cost function parameters in the A* algorithm to reflect human driving behavior patterns. Specifically, it adjusts the weighting factors for different cost components (turning costs, gear change costs, distance costs) to optimize the path planning output. This resolves the contradiction by changing the algorithm parameters to achieve more efficient parking paths without increasing system complexity
Solution Approach 2:
The patent copies human driving behavior patterns into the autonomous parking system by observing and replicating how human drivers naturally navigate parking spots. The system learns from human driving data to identify optimal path characteristics, such as minimizing unnecessary turns and gear changes. This copying approach improves parking efficiency while keeping the path planning algorithm relatively simple
2Extent of automation
If autonomous valet parking is implemented, then the system can autonomously park and retrieve vehicles, but the path planning must account for multiple constraints including obstacles, safety margins, and vehicle dynamics
Solution Approach 1:
The patent segments the path planning problem into distinct cost components that can be calculated and optimized separately. The total cost function is divided into multiple parts: turning costs, gear change costs, distance costs, and safety margin costs. Each component is handled by a separate calculation module, making the overall complex autonomous parking system manageable through modular decomposition
Solution Approach 2:
The patent introduces a cost function as an intermediary layer between the autonomous parking system and the path planning algorithm. This cost function acts as a mediator that translates multiple constraints (obstacles, safety margins, vehicle dynamics) into a unified optimization criterion. By using this intermediary, the system can handle complex constraints without directly complicating the path planning algorithm structure
3Speed
If the path planning optimizes for direct routes, then the parking speed increases, but the vehicle may encounter obstacles or violate safety margins
Solution Approach 1:
The patent applies equipotentiality by incorporating safety margins and obstacle avoidance constraints directly into the cost function, creating zones of equal safety potential around obstacles. The path planning algorithm naturally avoids these zones without requiring separate collision detection and correction steps. This approach maintains high parking speed while ensuring reliability, as the safe path is determined during the initial planning phase rather than through continuous correction
Data Source
AI summary
A method for autonomously parking or un-parking a motor vehicle includes locating the motor vehicle relative to a parking area, selecting a destination location within the parking area, generating a path from the location of the motor vehicle to the destination location, wherein the path includes a plurality of linked nodes, each node having a cost associated therewith, wherein the cost of a child node is equal to an inherited cost plus a base cost and a change cost, wherein the change cost is a function of characteristics of a parent node, and autonomously driving the motor vehicle along the path from the location of the motor vehicle to the destination location.