Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Data-Driven Predictive Planning and Control for Aerial 3D Inspection with Back-face Elimination

Savvas Papaioannou, Panayiotis Kolios, Christos G. Panayiotou, Marios M. Polycarpou

TL;DR

The paper addresses autonomous 3D inspection with UASs by unifying perception, planning, and control in a data-driven predictive framework that relies on input-output data rather than explicit models. It integrates back-face elimination into the planning loop to enable online visibility-aware, long-horizon trajectories, formulated as a MIQP (P1) that encodes inspection dynamics, perception constraints, and collision avoidance over a horizon $N$ while using a past/initiation window $K$ and Hankel-data representations. The approach uses a camera model with a convex pyramid FOV and orientation set to determine visible facets, and it demonstrates that back-face elimination improves visibility-based planning, enabling inspection of all facets even with large FOVs in simulation. Overall, the method provides a practical path toward real-time, perception-guided autonomous 3D inspection on off-the-shelf UAS platforms, with promising implications for scalable, coverage-focused inspection tasks.

Abstract

Automated inspection with Unmanned Aerial Systems (UASs) is a transformative capability set to revolutionize various application domains. However, this task is inherently complex, as it demands the seamless integration of perception, planning, and control which existing approaches often treat separately. Moreover, it requires accurate long-horizon planning to predict action sequences, in contrast to many current techniques, which tend to be myopic. To overcome these limitations, we propose a 3D inspection approach that unifies perception, planning, and control within a single data-driven predictive control framework. Unlike traditional methods that rely on known UAS dynamic models, our approach requires only input-output data, making it easily applicable to off-the-shelf black-box UASs. Our method incorporates back-face elimination, a visibility determination technique from 3D computer graphics, directly into the control loop, thereby enabling the online generation of accurate, long-horizon 3D inspection trajectories.

Data-Driven Predictive Planning and Control for Aerial 3D Inspection with Back-face Elimination

TL;DR

The paper addresses autonomous 3D inspection with UASs by unifying perception, planning, and control in a data-driven predictive framework that relies on input-output data rather than explicit models. It integrates back-face elimination into the planning loop to enable online visibility-aware, long-horizon trajectories, formulated as a MIQP (P1) that encodes inspection dynamics, perception constraints, and collision avoidance over a horizon while using a past/initiation window and Hankel-data representations. The approach uses a camera model with a convex pyramid FOV and orientation set to determine visible facets, and it demonstrates that back-face elimination improves visibility-based planning, enabling inspection of all facets even with large FOVs in simulation. Overall, the method provides a practical path toward real-time, perception-guided autonomous 3D inspection on off-the-shelf UAS platforms, with promising implications for scalable, coverage-focused inspection tasks.

Abstract

Automated inspection with Unmanned Aerial Systems (UASs) is a transformative capability set to revolutionize various application domains. However, this task is inherently complex, as it demands the seamless integration of perception, planning, and control which existing approaches often treat separately. Moreover, it requires accurate long-horizon planning to predict action sequences, in contrast to many current techniques, which tend to be myopic. To overcome these limitations, we propose a 3D inspection approach that unifies perception, planning, and control within a single data-driven predictive control framework. Unlike traditional methods that rely on known UAS dynamic models, our approach requires only input-output data, making it easily applicable to off-the-shelf black-box UASs. Our method incorporates back-face elimination, a visibility determination technique from 3D computer graphics, directly into the control loop, thereby enabling the online generation of accurate, long-horizon 3D inspection trajectories.

Paper Structure

This paper contains 14 sections, 1 theorem, 11 equations, 2 figures, 1 algorithm.

Table of Contents

  1. Introduction
  2. Related Work
  3. Problem Formulation
  4. UAS Dynamical Model
  5. UAS Camera Model
  6. Problem Statement
  7. Planning and Control for 3D Inspection
  8. Inspection Planning Dynamics
  9. Back-face Elimination
  10. Data-Driven Predictive Inspection Control
  11. Evaluation
  12. Simulation Setup
  13. Results
  14. Conclusion

Key Result

Lemma 1

Consider a controllable LTI system $\mathscr{B}(A,B,C,D)$ described by Eq. eq:LTI and assume that the input sequence (i.e., control input samples) $u^d \in \mathbb{R}^{m \times T}$ is persistently exciting of order $N+n$, and $y^d \in \mathbb{R}^{p \times T}$ is the corresponding output (i.e., colle Subsequently, a valid input-output trajectory $(u, y)$ of the system of length $N$ can be construct

Figures (2)

  • Figure 1: The figure provides an example to illustrate the proposed approach of data-driven predictive planning and control for 3D inspection.
  • Figure 2: The effect of back-face elimination (BFE) on the performance of 3D inspection planning. Average percentage of visible inspected facets at the end of the mission as a function of the FOV size.

Theorems & Definitions (2)

  • Definition 1
  • Lemma 1: Fundamental LemmaMarkovsky2008data