Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

An Intent Modeling and Inference Framework for Autonomous and Remotely Piloted Aerial Systems

Kesav Kaza, Varun Mehta, Hamid Azad, Miodrag Bolic, Iraj Mantegh

TL;DR

The paper addresses protecting geo-fences from unauthorized UAVs by introducing an integrated intent modeling framework based on a novel mathematical definition of intent via critical waypoint patterns (CWPP) and motion processes. It couples a mission-planner perspective with a defense-planner viewpoint, leveraging multi-mode kinematics and an IMM filter to extract features for intent inference. An attention-based Bi-LSTM classifier is trained within this framework, and the approach is demonstrated through 2D/3D simulations using RRT* path planning and radar data to show trajectory generation, tracking, and intent recognition performance. The work provides loss-function formulations for classification, probabilistic and time-constrained protection objectives, and highlights how synthetic datasets with labeled intents can be used to train robust intent classifiers, contributing to proactive geo-fence protection and UAV security research.

Abstract

An intent modelling and inference framework is presented to assist the defense planning for protecting a geo-fence against unauthorized flights. First, a novel mathematical definition for the intent of an uncrewed aircraft system (UAS) is presented. The concepts of critical waypoints and critical waypoint patterns are introduced and associated with a motion process to fully characterize an intent. This modelling framework consists of representations of a UAS mission planner, used to plan the aircraft's motion sequence, as well as a defense planner, defined to protect the geo-fence. It is applicable to autonomous, semi-autonomous, and piloted systems in 2D and 3D environments with obstacles. The framework is illustrated by defining a library of intents for a security application. Detection and tracking of the target are presumed for formulating the intent inference problem. Multiple formulations of the decision maker's objective are discussed as part of a deep-learning-based methodology. Further, a multi-modal dynamic model for characterizing the UAS flight is discussed. This is later utilized to extract features using the interacting multiple model (IMM) filter for training the intent classifier. Finally, as part of the simulation study, an attention-based bi-directional long short-term memory (Bi-LSTM) network for intent inference is presented. The simulation experiments illustrate various aspects of the framework, including trajectory generation, radar measurement simulation, etc., in 2D and 3D environments.

An Intent Modeling and Inference Framework for Autonomous and Remotely Piloted Aerial Systems

TL;DR

The paper addresses protecting geo-fences from unauthorized UAVs by introducing an integrated intent modeling framework based on a novel mathematical definition of intent via critical waypoint patterns (CWPP) and motion processes. It couples a mission-planner perspective with a defense-planner viewpoint, leveraging multi-mode kinematics and an IMM filter to extract features for intent inference. An attention-based Bi-LSTM classifier is trained within this framework, and the approach is demonstrated through 2D/3D simulations using RRT* path planning and radar data to show trajectory generation, tracking, and intent recognition performance. The work provides loss-function formulations for classification, probabilistic and time-constrained protection objectives, and highlights how synthetic datasets with labeled intents can be used to train robust intent classifiers, contributing to proactive geo-fence protection and UAV security research.

Abstract

An intent modelling and inference framework is presented to assist the defense planning for protecting a geo-fence against unauthorized flights. First, a novel mathematical definition for the intent of an uncrewed aircraft system (UAS) is presented. The concepts of critical waypoints and critical waypoint patterns are introduced and associated with a motion process to fully characterize an intent. This modelling framework consists of representations of a UAS mission planner, used to plan the aircraft's motion sequence, as well as a defense planner, defined to protect the geo-fence. It is applicable to autonomous, semi-autonomous, and piloted systems in 2D and 3D environments with obstacles. The framework is illustrated by defining a library of intents for a security application. Detection and tracking of the target are presumed for formulating the intent inference problem. Multiple formulations of the decision maker's objective are discussed as part of a deep-learning-based methodology. Further, a multi-modal dynamic model for characterizing the UAS flight is discussed. This is later utilized to extract features using the interacting multiple model (IMM) filter for training the intent classifier. Finally, as part of the simulation study, an attention-based bi-directional long short-term memory (Bi-LSTM) network for intent inference is presented. The simulation experiments illustrate various aspects of the framework, including trajectory generation, radar measurement simulation, etc., in 2D and 3D environments.
Paper Structure (13 sections, 10 equations, 11 figures, 3 tables)

This paper contains 13 sections, 10 equations, 11 figures, 3 tables.

Table of Contents

  1. Introduction
  2. System Model
  3. Intent Models
  4. Kinematic motion model
  5. Representation of Objectives
  6. Intent inference as multi-class classification problem
  7. Multi-class classification with class imbalance and asymmetric error costs
  8. Probabilistic protection requirement
  9. Time-constrained protection requirement
  10. Simulation Experiments
  11. Classifier architecture
  12. Radar measurements and non-optimal paths
  13. Conclusion

Figures (11)

  • Figure 1: The intent inference framework with a representation of the mission planner and the defence planner. The mission planner drives a UAV in the environment with a hidden intent, and the defence planner infers this intent using the specified modules. In general, the mission planner's representation is from the perspective of the defence planner.
  • Figure 3: The various modules of the simulator. RRT* algorithm was used for path finding. The UAV trajectory and radar measurement simulation was implemented with MATLAB UAV and Radar toolbox.
  • Figure 4: [left] Architecture of the intent classifier. [right] Evolution of training and validation accuracy with number of epochs for various splits.
  • Figure 5: Sample trajectories on 2D and 3D environments for various intents generated by the path finding algorithm RRT*.
  • Figure 6: Examples of trajectories along with the simulated radar tracks with a range error which falls within $2$ meters once the tracking filter is stabilized.
  • ...and 6 more figures

Theorems & Definitions (2)

  • Definition 1
  • Definition 2