AERIAL-CORE: AI-Powered Aerial Robots for Inspection and Maintenance of Electrical Power Infrastructures

Aníbal Ollero, Alejandro Suárez, Christos Papaioannidis, Ioannis Pitas, Juan M. Marredo, Viet Duong Hoang, Emad Ebeid, Vít Krátký, Martin Saska, Chloe Hanoune, Amr Afifi, Antonio Franchi, Charalampos Vourtsis, Dario Floreano, Goran Vasiljevic, Stjepan Bogdan, Alvaro Caballero, Fabio Ruggiero, Vincenzo Lippiello, Carlos Matilla, Giovanni Cioffi, Davide Scaramuzza, José Ramiro Martinez de Dios, Begoña C. Arrue, Carlos Martín, Krzysztof Zurad, Carlos Gaitan, Jacob Rodriguez, Antonio Munoz, Antidio ViguriaarXiv 2024 · DOI · arXiv

Introduction

Maintaining global electrical grids is a dangerous, labor-intensive, and costly endeavor. Traditionally, utility companies have relied on manned helicopters or human workers climbing high-voltage towers at extreme heights. The AERIAL-CORE project introduces a paradigm shift: the first fully integrated, autonomous ecosystem of aerial robots designed to manage the complete lifecycle of power line Inspection and Maintenance (I&M).

The AERIAL-CORE system architecture, illustrating the workflow from mission request to results across three specialized teams.

Addressing the Gaps in Power Line Maintenance

While drones are increasingly used in the power sector, current applications are often limited to visual inspection within a pilot's line of sight. AERIAL-CORE addresses three critical challenges that have previously hindered full automation:

  1. Range and Efficiency: Developing drones that combine the high-speed transit of fixed-wing aircraft with the precision hovering of multi-rotors.
  2. Physical Manipulation: Enabling robots to perform hardware installation (e.g., bird diverters) while hovering in high-wind, high-voltage environments.
  3. Human-Robot Collaboration: Providing robotic "co-workers" that enhance the safety of human technicians on-site by monitoring safety distances and delivering tools.

Technological Pillars

The system relies on several innovative technological pillars to achieve its goals:

Morphing Airframes

The project utilizes specialized UAVs such as Marvin-5, a tilt-rotor that adjusts its wingspan (from 1.85m to 2.41m) for optimal efficiency during transit or data capture. Another innovation is Morpho, a tailsitter VTOL with four independently actuated wings that can fold to sustain higher wind speeds (up to 9 m/s) during precision hovering.

Perching and Manipulation

A key innovation is the shift from hovering to perching. The Main Local Manipulation Platform (MLMP), a 45kg robot, perches directly on the power line. This converts a complex dynamic control problem into a static one, allowing for high-precision maintenance tasks without the battery drain of active flight. The platform features a 6-DOF anthropomorphic arm capable of handling 5 kg loads.

Advanced Perception and Energy Harvesting

The system employs a combination of LIDAR, standard RGB cameras, and Event Cameras—which are robust to motion blur—for 3D semantic mapping of towers and lines. To solve the range problem, drones can install charging stations that harvest energy directly from the power line's magnetic field, showing a 58.6% increase in harvested power over traditional methods.

Data-Driven Intelligence

The project utilized several datasets to train its AI components. A synthetic dataset of ~30,000 simulated images was created for power line detection, enabling "zero-shot" transfer to real-world environments. For human-robot interaction, the system was trained on the open-multidrone-datasets and auth-uav-gesture-dataset, allowing the robots to recognize gestures from workers in safety gear.

Key Results and Field Validation

The AERIAL-CORE system was validated in real-world conditions at the ATLAS experimental flight center in Spain, covering over 1,000 km² of airspace. Key achievements include:

  • BVLOS Operations: Successful autonomous flights and inspections over 10 km from ground control stations.
  • Rapid Mapping: Mapping 10 km of power lines in just 10 minutes using real-time LIDAR-based semantic mapping.
  • Autonomous Maintenance: The system successfully demonstrated the autonomous installation of bird diverters and recharging stations on live lines.
Autonomous deployment and installation of maintenance devices on power lines using dual-arm aerial manipulators.

Conclusion and Future Outlook

While AERIAL-CORE demonstrates the technical feasibility of autonomous grid maintenance, regulatory hurdles for Beyond Visual Line of Sight (BVLOS) operations remain a significant bottleneck. Future research will focus on establishing "drone highways" over power lines and expanding the repertoire of maintenance tasks, such as insulator cleaning and cable repair, to move toward a permanent robotic presence on the electrical grid.

Summary updated

BibTeX
@misc{ollero2024aerial,
  author = {Aníbal Ollero and Alejandro Suárez and Christos Papaioannidis and Ioannis Pitas and Juan M. Marredo and Viet Duong Hoang and Emad Ebeid and Vít Krátký and Martin Saska and Chloe Hanoune and Amr Afifi and Antonio Franchi and Charalampos Vourtsis and Dario Floreano and Goran Vasiljevic and Stjepan Bogdan and Alvaro Caballero and Fabio Ruggiero and Vincenzo Lippiello and Carlos Matilla and Giovanni Cioffi and Davide Scaramuzza and José Ramiro Martinez de Dios and Begoña C. Arrue and Carlos Martín and Krzysztof Zurad and Carlos Gaitan and Jacob Rodriguez and Antonio Munoz and Antidio Viguria},
  title = {{AERIAL-CORE: AI-Powered Aerial Robots for Inspection and Maintenance of Electrical Power Infrastructures}},
  year = {2024},
  eprint = {2401.02343},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2401.02343},
}
Figure