Vehicle Predictive Optimization With Safe-State Horizon Constraints

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing predictive optimization methods in vehicles are computationally intense and use is limited to a relatively short prediction horizon, which have not been addressed in the field of environmental pollution control and purification.

Innovation Solution

The use of predictive optimization methods are used to a relatively short prediction horizon, which have not been addressed in the field of environmental pollution control and purification.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If predictive optimization is used with a longer prediction horizon, then the ability to plan for safe vehicle states is improved, but computational intensity increases making it infeasible for real-time control

Engineering Contradiction:
Improveability to ensure safe vehicle statesVSAvoidcomputational intensity
Core Design Contradiction:
ReliabilityVSPower

Solution Approach 1:

The prediction horizon is segmented into multiple shorter sub-horizons. The optimization problem is divided into sequential sub-problems, each solving for a shorter time window. This segmentation reduces the computational complexity of each individual optimization problem while collectively covering the longer prediction horizon needed for safety assurance.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

Feasibility checks and constraints are prepared in advance based on the longer prediction horizon requirements. The system pre-computes safe state regions and feasibility conditions that must be satisfied at each sub-horizon step, allowing the real-time optimizer to work with pre-defined safety boundaries rather than computing everything from scratch.

Inventive Principle:
Principle #10Preliminary action

2Reliability

If predictive optimization is constrained by safety requirements, then the risk of unfeasible vehicle states is reduced, but the feasible region for optimization is limited

Engineering Contradiction:
Improverisk reduction of unfeasible vehicle statesVSAvoidfeasible region for optimization
Core Design Contradiction:
ReliabilityVSAdaptability or versatility

Solution Approach 1:

The feasible region boundaries are made dynamic rather than static. The optimization constraints adapt based on current vehicle state, environmental conditions, and predicted future states. This allows the feasible region to expand or contract dynamically, providing more flexibility when conditions permit while maintaining safety boundaries when risks are present.

Inventive Principle:
Principle #15Dynamics

Solution Approach 2:

The system continuously monitors the vehicle state and optimization results, using feedback to adjust the feasible region constraints. When the optimizer approaches boundary conditions or when safety margins are reduced, feedback mechanisms modify the constraints to prevent unfeasible states while allowing maximum optimization freedom when safety is not at risk.

Inventive Principle:
Principle #23Feedback

Data Source

PatentEP4667313A1Determining vehicle control parameters using predictive optimization with enhanced constraint
Publication Date: 2025.12.24 VOLVO TRUCK CORP
  • EP4667313A1 patent drawingFigure 1~3
  • EP4667313A1 patent drawingFigure 4~6B
  • EP4667313A1 patent drawingFigure 7~9

AI summary

A computer system (15) comprising processing circuitry (21) configured to: determine a set of vehicle control parameters, using predictive optimization (29) of a vehicle model with a first prediction horizon (h1), the predictive optimization (29) being constrained by a set of allowed vehicle states at the first prediction horizon (h1); and control an operation of a vehicle (1) using the determined set of vehicle control parameters, wherein: the set of allowed vehicle states at the first prediction horizon (h1) is a set of initial vehicle states (S0) estimated to result in a set of safe vehicle states (Sh2) at a second prediction horizon (h2) longer than the first prediction horizon (h1).