Formalizing the Execution Context of Behavior Trees for Runtime Verification of Deliberative Policies

TL;DR

This paper presents a formal approach to verify robot behavior by formalizing Behavior Trees' execution context, enabling runtime verification to prevent unexpected behaviors in robotic control systems.

Contribution

It introduces a formal property specification language and a message-passing model for Behavior Trees, facilitating scalable runtime verification in robotic architectures.

Findings

Feasibility demonstrated on real robots and simulations.

Effective integration with robotic software architectures.

Provides an OS-level virtualization environment for reproducibility.

Abstract

In this paper, we enable automated property verification of deliberative components in robot control architectures. We focus on formalizing the execution context of Behavior Trees (BTs) to provide a scalable, yet formally grounded, methodology to enable runtime verification and prevent unexpected robot behaviors. To this end, we consider a message-passing model that accommodates both synchronous and asynchronous composition of parallel components, in which BTs and other components execute and interact according to the communication patterns commonly adopted in robotic software architectures. We introduce a formal property specification language to encode requirements and build runtime monitors. We performed a set of experiments, both on simulations and on the real robot, demonstrating the feasibility of our approach in a realistic application and its integration in a typical robot software architecture. We also provide an OS-level virtualization environment to reproduce the experiments in the simulated scenario.