Cost-Based Path Planning for Autonomous Vehicles
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous driving vehicles face challenges in optimizing motion planning and control without initial constraints, leading to inaccurate and rough navigation under various circumstances.
Innovation Solution
The implementation of a decision and planning system that uses quadratic programming and gradient descent optimization to calculate optimal trajectories for autonomous driving vehicles, considering traffic rules and obstacle information, by generating path and speed profiles that minimize total costs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If motion planning and control are applied without initial constraints, then the system has high adaptability to various circumstances, but the navigation accuracy and smoothness deteriorate
Solution Approach 1:
The system performs preliminary action by pre-calculating cost functions and constraints before actual motion planning. The cost function framework is established in advance with predefined parameters for path smoothness, collision avoidance, and energy efficiency, allowing the planner to make accurate decisions without needing to adapt to each circumstance from scratch.
Solution Approach 2:
The system applies parameter changes by dynamically adjusting the weights and parameters within the cost function based on different driving scenarios. The optimization algorithm modifies trajectory parameters such as position, velocity, and acceleration to achieve smooth and accurate navigation while adapting to various environmental conditions.
2Adaptability or versatility
If optimization is performed without initial constraints, then the system can handle diverse scenarios, but the computational complexity and time consumption increase
Solution Approach 1:
The system uses parameter changes by implementing a layered optimization approach where coarse-level constraints are applied first to reduce the search space, followed by fine-level optimization. This allows the system to handle diverse scenarios efficiently by adjusting the granularity of optimization based on scenario complexity.
Solution Approach 2:
The system applies preliminary action by pre-defining constraint frameworks and cost function structures that can be quickly instantiated for different scenarios. This preliminary setup reduces computational overhead during real-time operation, allowing the system to handle diverse scenarios without excessive time consumption.
3Adaptability or versatility
If general motion planning is used for all vehicle types, then the system has high versatility, but the path smoothness and accuracy deteriorate under specific circumstances
Solution Approach 1:
The system applies local quality by customizing the cost function parameters and constraints according to specific vehicle characteristics and operational requirements. Different vehicle types can have tailored smoothness parameters, acceleration limits, and path constraints, ensuring optimal path quality for each vehicle while maintaining the overall versatility of the planning framework.
Data Source
AI summary
According to some embodiments, a system generates a number of possible decisions for routing the ADV from a first location to a second location based on perception information perceiving a driving environment surrounding the ADV, including one or more obstacles in view of a set of traffic rules. The system calculates a number of trajectories based on a combination of one or more of the possible decisions. The system calculates a total cost for each of the trajectories using a number of cost functions and selects one of the trajectories with a minimum total cost as the driving trajectory to control the ADV autonomously. The cost functions include a path cost function, a speed cost function, and an obstacle cost function.