DDP Planning Architecture for Autonomous Vehicle Path Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current optimization architectures for autonomous driving vehicles, such as Sequential Quadratic Programming (SQP) and Quadratic Programming (QP), are inefficient as they optimize a single candidate path at a time, leading to increased processing time for determining an optimal path trajectory.
Innovation Solution
A Differential Dynamic Programming (DDP) based planning architecture that optimizes multiple candidate paths in parallel, allowing for the selection of the most optimal path for an autonomous driving vehicle to navigate, while relaxing constraints if necessary to satisfy path constraints.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If Sequential Quadratic Programming (SQP) or Quadratic Programming (QP) is used to optimize a single candidate path at a time, then the optimization process is simple and manageable, but the processing time increases significantly
Solution Approach 1:
The patent divides the path optimization problem into multiple independent candidate paths that can be processed in parallel. Instead of optimizing one path sequentially, the system generates multiple candidate paths and optimizes them simultaneously using separate processor cores, thereby reducing total processing time while maintaining optimization quality
Solution Approach 2:
The patent transitions from single-core sequential optimization to multi-core parallel optimization by adding a dimensional aspect of concurrent processing. Multiple processor cores are utilized to optimize multiple candidate paths at the same time, effectively transforming the processing architecture from one-dimensional sequential to multi-dimensional parallel execution
2Loss of time
If multiple candidate paths are optimized simultaneously, then the processing time is reduced, but the device complexity increases due to multi-core processing requirements
Solution Approach 1:
The patent creates a universal optimization framework that can handle multiple candidate paths using the same optimization algorithm and constraints. The multi-core processor architecture is designed to execute identical optimization routines in parallel, allowing the system to maintain algorithmic simplicity while achieving speedup through parallelization
Solution Approach 2:
The patent replicates the optimization process across multiple processor cores, creating copies of the optimization algorithm that run independently on different candidate paths. Each core maintains a complete copy of the optimization logic, enabling true parallel execution without complex inter-core communication requirements
Data Source
AI summary
In one embodiment, method performed by an autonomous driving vehicle (ADV) that determines, within a driving space, a plurality of routes from a current location of the ADV to a desired location. The method determines, for each route of the plurality of routes, an objective function to control the ADV autonomously along the route and, for each of the objective functions, performs Differential Dynamic Programming (DDP) optimization in view of a set of constraints to produce a path trajectory. The method determines whether at least one of the path trajectories satisfies each constraint and, in response to a path trajectory satisfying each of the constraints, selects the path trajectory for navigating the ADV from the current location to the desired location.