Behavior Tree Translation to Petri Nets for Formal Verification

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Behavior trees, commonly used in robotics for designing control policies, lack a universally accepted formal definition and associated verification methods, making formal verification of autonomous systems difficult or impossible.

Innovation Solution

A method is introduced to translate behavior trees into Petri nets, enabling formal verification by using state-of-the-art model checkers to verify temporal logic properties, with optional optimizations to reduce computational complexity.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of operation

If behavior trees are used to implement autonomous system control, then the system gains simplicity and ease of operation, but formal verification capability is lost

Engineering Contradiction:
Improveease of operationVSAvoidformal verification capability
Core Design Contradiction:
Ease of operationVSReliability

Solution Approach 1:

The patent creates a formal copy of the behavior tree in the form of a Petri net model. The translation process generates an equivalent Petri net that preserves the behavioral semantics of the original behavior tree, enabling formal verification while maintaining the original implementation's simplicity. The Petri net serves as a formal representation that can be verified independently without modifying the actual behavior tree code.

Inventive Principle:
Principle #26Copying

2Reliability

If behavior trees are translated to Petri nets for formal verification, then verification capability is improved, but computational complexity increases

Engineering Contradiction:
Improveformal verification capabilityVSAvoidcomputational complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The patent applies segmentation by translating the behavior tree into a hierarchical Petri net structure that mirrors the tree's modular organization. Each behavior tree node is translated into corresponding Petri net components (places, transitions, arcs), allowing the verification process to work with structured, manageable segments rather than a monolithic complex model. This hierarchical approach reduces the computational burden by enabling localized analysis and optimization.

Inventive Principle:
Principle #1Segmentation

3Measurement precision

If exhaustive model checking is performed on the Petri net, then verification completeness is improved, but computational resources and time are excessive

Engineering Contradiction:
Improveverification completenessVSAvoidverification time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The patent implements partial verification by allowing users to specify subsets of properties, states, or transitions to verify based on their priorities and resource constraints. Rather than requiring complete exhaustive model checking of the entire Petri net, the system enables verification of critical properties first, providing progressively deeper analysis as needed. This approach delivers useful verification results faster while maintaining the option for more comprehensive checking when resources permit.

Inventive Principle:
Principle #16Partial or excessive action

Data Source

PatentEP4614313A1Translating behavior trees to petri nets for model checking
Publication Date: 2025.09.10 ROBERT BOSCH GMBH
  • EP4614313A1 patent drawingFigure 1
  • EP4614313A1 patent drawingFigure 2a
  • EP4614313A1 patent drawingFigure 2b

AI summary

A computer-implemented method for verifying the autonomous behavior of an at least partially autonomous system, wherein the autonomous behavior is based on an implementation of a behavior tree, the method comprising: translating the behavior tree, or another representation thereof, to a Petri net; checking the autonomous behavior based on the Petri net, thereby generating a test result.