Autonomous Driving Speed Planning on ST Graphs With MPQP
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for autonomous driving trajectory planning struggle with non-convex issues, computational inefficiencies, and challenges in synchronizing space-time dynamics, especially when dealing with dynamic obstacles, leading to suboptimal solutions and potential local optima.
Innovation Solution
A method using Multi-Profile Quadratic Programming (MPQP) that decouples path and speed planning, constructs a space-time (ST) graph, segments it into cells, and employs Breadth-First Search (BFS) to identify viable paths, integrating these into quadratic programming for optimal speed profiles while considering dynamic obstacles and kinematic limits.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If iterative QP approximations are used to solve non-convex problems on ST graph, then space-time synchronization is improved, but computational complexity increases significantly
Solution Approach 1:
The patent segments the continuous speed planning problem into discrete speed profiles (e.g., aggressive, moderate, conservative) that can be evaluated independently. Each profile represents a predefined speed pattern that satisfies kinematic constraints, allowing the system to select optimal profiles without iterative approximation while maintaining space-time synchronization.
Solution Approach 2:
The patent introduces dynamic obstacle ordering constraints that adaptively adjust the relative timing of obstacle crossings based on current traffic conditions. This dynamic constraint adjustment enables space-time synchronization without requiring computationally heavy iterative QP solutions, as the constraints are formulated to be directly enforceable in the optimization framework.
2Reliability
If Dynamic Programming is used to optimize non-convex problems on ST graph, then global optimality is improved, but dimensionality limits the number of states and control inputs
Solution Approach 1:
The patent extracts the speed planning decision from the full trajectory optimization problem by decoupling path and speed planning. This extraction reduces the dimensionality of the optimization problem by focusing only on speed profile selection rather than simultaneous path and speed optimization, enabling global optimality without DP's dimensional limitations.
Solution Approach 2:
The patent transforms the continuous speed control problem into a discrete parameter selection problem by defining a finite set of speed profiles with distinct characteristics. This parameter transformation allows the system to achieve global optimality through discrete profile selection rather than continuous optimization, avoiding DP's dimensionality constraints while maintaining optimality guarantees.
3Ease of manufacture
If space domain planning is used to integrate position-based speed limits, then ease of integration is improved, but zero speed handling becomes problematic with infinity time evaluation
Solution Approach 1:
The patent transitions from space-domain planning to time-domain planning by formulating the optimization in the temporal dimension. This dimensional shift allows natural handling of zero speed scenarios through time parameterization, where stop durations are explicitly modeled as time intervals rather than causing infinity evaluations. Position-based speed limits are integrated through time-synchronized constraints rather than spatial constraints.
Data Source
AI summary
A method for generating operable driving areas for an autonomous driving vehicle based on a path trajectory of the autonomous driving vehicle is provided. The method may form a space time (ST) graph indicating a distance of travel along the path trajectory with respect to time of the autonomous driving vehicle and path trajectories of devices intersecting with the path trajectory of the autonomous driving vehicle. The method may segment the ST graph into cells, wherein viable cells represent discretized viable unoccupied spaces in the ST graph. The method may find passage ways for the autonomous driving vehicle based on the viable cells. The method may select a desired passage way using quadratic programming (QP) optimization when multiple passage ways are found.