Vehicle Trajectory Planning With QP Linearization and Heuristics

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing trajectory planning methods for autonomous vehicles face challenges in efficiently solving complex driving scenarios within a limited timeframe, often getting stuck at local minima and requiring increased computational effort due to separate longitudinal and lateral planning, and lacking heuristic evaluation capabilities.

Innovation Solution

A method that integrates domain-specific heuristics into an SQP-based solver to split the non-convex and non-linear trajectory planning problem into sequential, locally accurate, convex quadratic programs, using a QP solver to optimize the trajectory planning by discretizing and linearizing environmental data at each operating point.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If nonlinear general-purpose solvers are used to solve trajectory planning problems, then the problem can be solved in a generic manner, but the computational effort increases and the solution may get stuck at local minima

Engineering Contradiction:
Improvegeneric problem solving capabilityVSAvoidcomputational efficiency
Core Design Contradiction:
Adaptability or versatilityVSProductivity

Solution Approach 1:

The patent transforms the nonlinear trajectory planning problem into a quadratic programming problem by changing the mathematical parameters and formulation. This allows using efficient QP solvers instead of general nonlinear solvers, achieving both speed and reliability without getting stuck at local minima

Inventive Principle:
Principle #35Parameter changes

Solution Approach 2:

The patent replaces the mechanical nonlinear optimization approach with a quadratic programming framework. By substituting the solution methodology from general nonlinear solvers to QP-based solvers, computational efficiency is dramatically improved while maintaining solution quality

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

2Ease of manufacture

If separate lateral and longitudinal trajectory planning is performed, then each dimension can be optimized independently, but the overall computational effort increases

Engineering Contradiction:
Improveindependent optimization capabilityVSAvoidcomputational effort
Core Design Contradiction:
Ease of manufactureVSProductivity

Solution Approach 1:

The patent merges separate lateral and longitudinal trajectory planning into a unified quadratic programming framework. By combining both dimensions in a single QP formulation, the computational effort is reduced compared to sequential optimization while maintaining the ability to independently constrain each dimension

Inventive Principle:
Principle #5Merging (Combining)

3Productivity

If boundary conditions are only adjusted between planning loops, then computational steps are reduced, but the accuracy of trajectory planning decreases

Engineering Contradiction:
Improvecomputational stepsVSAvoidtrajectory planning accuracy
Core Design Contradiction:
ProductivityVSManufacturing precision

Solution Approach 1:

The patent ensures continuous adjustment of boundary conditions at each operating point along the trajectory rather than only between planning loops. This continuous update approach maintains high accuracy while working within the QP framework efficiently

Inventive Principle:
Principle #20Continuity of useful action

Data Source

PatentEP4412882B1Method for planning a trajectory of a driving maneuver of a motor vehicle, computer program product, computer-readable storage medium, and vehicle
Publication Date: 2025.11.05 AUMOVIO AUTONOMOUS MOBILITY GERMANY GMBH
  • EP4412882B1 patent drawingFigure 1~2
  • EP4412882B1 patent drawingFigure 3~4

AI summary

The invention relates to a method for planning a trajectory of a driving maneuver of a motor vehicle (4), said method having the steps of: 1) evaluating (200) a current driving situation; 2) ascertaining an adapted driving maneuver on the basis of the evaluated driving situation; 3) calculating a number of working points (A1-A4) of the trajectory on the basis of the evaluated driving situation and the ascertained driving maneuver; 4) discretizing the current surroundings information and/or vehicle information at each of the working points (A1-A4) of the trajectory; 5) selecting and integrating (300) relevant surroundings data for each working point (A1-A4) on the basis of domain-specific information; 6) linearizing the selected and integrated surroundings data; 7) using the linearized surroundings data in order to formulate a QP model; 8) solving (400) the QP model using a QP solver; and 9) repeating (500) steps 1) to 8) on the basis of a convergence of the solution, wherein upon repeating steps 1) to 8), the solution of the QP solver is taken into consideration.