Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

High-Level Multi-Robot Trajectory Planning And Spurious Behavior Detection

Fernando Salanova, Jesús Roche, Cristian Mahuela, Eduardo Montijano

TL;DR

This paper tackles spurious behaviors in multi-robot missions specified by LTL by coupling Nets-with-Nets formal specifications with a Transformer-based anomaly detector trained on action-token embeddings derived from NWN-generated trajectories. The methodology comprises five stages: NWN-based formalization, Renew-driven trajectory generation (with spurious data augmentation), VMAS-based low-level timing, a rich trajectory-embedding scheme, and a Transformer classifier trained per anomaly type. Results show high accuracy for explicit mission violations (e.g., near-perfect for Forbidden Zone), strong performance on temporal-order violations, and robust detection across several anomaly classes, with ablation confirming the importance of the encoding and architecture choices. The work enables end-to-end detection of mission-level spuriousness in heterogeneous multi-robot teams, offering a practical pathway to verify high-level mission integrity in complex environments.

Abstract

The reliable execution of high-level missions in multi-robot systems with heterogeneous agents, requires robust methods for detecting spurious behaviors. In this paper, we address the challenge of identifying spurious executions of plans specified as a Linear Temporal Logic (LTL) formula, as incorrect task sequences, violations of spatial constraints, timing inconsis- tencies, or deviations from intended mission semantics. To tackle this, we introduce a structured data generation framework based on the Nets-within-Nets (NWN) paradigm, which coordinates robot actions with LTL-derived global mission specifications. We further propose a Transformer-based anomaly detection pipeline that classifies robot trajectories as normal or anomalous. Experi- mental evaluations show that our method achieves high accuracy (91.3%) in identifying execution inefficiencies, and demonstrates robust detection capabilities for core mission violations (88.3%) and constraint-based adaptive anomalies (66.8%). An ablation experiment of the embedding and architecture was carried out, obtaining successful results where our novel proposition performs better than simpler representations.

High-Level Multi-Robot Trajectory Planning And Spurious Behavior Detection

TL;DR

This paper tackles spurious behaviors in multi-robot missions specified by LTL by coupling Nets-with-Nets formal specifications with a Transformer-based anomaly detector trained on action-token embeddings derived from NWN-generated trajectories. The methodology comprises five stages: NWN-based formalization, Renew-driven trajectory generation (with spurious data augmentation), VMAS-based low-level timing, a rich trajectory-embedding scheme, and a Transformer classifier trained per anomaly type. Results show high accuracy for explicit mission violations (e.g., near-perfect for Forbidden Zone), strong performance on temporal-order violations, and robust detection across several anomaly classes, with ablation confirming the importance of the encoding and architecture choices. The work enables end-to-end detection of mission-level spuriousness in heterogeneous multi-robot teams, offering a practical pathway to verify high-level mission integrity in complex environments.

Abstract

The reliable execution of high-level missions in multi-robot systems with heterogeneous agents, requires robust methods for detecting spurious behaviors. In this paper, we address the challenge of identifying spurious executions of plans specified as a Linear Temporal Logic (LTL) formula, as incorrect task sequences, violations of spatial constraints, timing inconsis- tencies, or deviations from intended mission semantics. To tackle this, we introduce a structured data generation framework based on the Nets-within-Nets (NWN) paradigm, which coordinates robot actions with LTL-derived global mission specifications. We further propose a Transformer-based anomaly detection pipeline that classifies robot trajectories as normal or anomalous. Experi- mental evaluations show that our method achieves high accuracy (91.3%) in identifying execution inefficiencies, and demonstrates robust detection capabilities for core mission violations (88.3%) and constraint-based adaptive anomalies (66.8%). An ablation experiment of the embedding and architecture was carried out, obtaining successful results where our novel proposition performs better than simpler representations.
Paper Structure (14 sections, 9 equations, 3 figures, 3 tables)

This paper contains 14 sections, 9 equations, 3 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Methodology
  4. Formal specification of multi-robot missions via NWNs
  5. Dataset generation: simulating trajectories with Renew
  6. Low-Level Simulation
  7. Trajectory Encoding for Machine Learning
  8. Machine Learning Architecture
  9. Implementation
  10. High-Level Behaviors and Anomalies
  11. Training and Evaluation
  12. Experimental Results and Analysis
  13. Limitations & Future work
  14. Conclusion

Figures (3)

  • Figure 1: Conceptual structure of the Nets-within-Nets framework. (i) Represents the Specification PN, in charge of ensuring a correct development of the mission. (ii) the System Net, that synchronizes the state evolution in respect to the mission and the robot actions. (iii) pictures the Robot PNs which models the agents environment and movement.
  • Figure 2: Visualization of designed scenario and low-level trajectories simulated in VMAS for the given example $\lozenge ( y_1 \land \lozenge ( y_2 \land \lozenge y_3 ) )$
  • Figure 3: Overview of the Transformer-based network architecture for spurious trajectory detection.(Embedding block explained in section.\ref{['ssec:encoding']})