RAVEN :Recognition of Asset and Vegetation for Electrical Network

[mohamedGaha1]

Mission Statement

The RAVEN (Recognition of Asset and Vegetation for Electrical Network) project is designed to democratize mobile LiDAR–based analytics for overhead electrical distribution networks by providing openly accessible algorithms capable of automatically detecting, modeling, and classifying power distribution assets as well as surrounding vegetation. The objectives of the project are to enhance grid reliability, enable the development of high‑fidelity digital twins, and accelerate data‑driven decision making within utilities through open, collaborative, and interoperable analytical solutions.

Description

RAVEN is an open source library designed to extract actionable intelligence from LiDAR point clouds for the distribution networks. It includes four main algorithms:

1. Mobile point cloud classifier for Power Network Assets
2. Power line overhead span extractor from classified point cloud
3. Highly obstructed power line modelling using mathematical reconstruction
4. Vegetation encroachment detection
   These modules address the challenge of incomplete or obstructed mobile LiDAR data by combining geometric modeling, AI algorithm and contextual information. RAVEN enables utilities to build accurate digital twins, improve asset mapping, and enhance vegetation management risk assessment.

Proposed Project Stage

Sandbox

Is this a new project or an existing one?

New Project

Current lead(s)

• Mohamed Gaha, Philippe Massicotte, Thomas Tolhurst : Researchers (CRHQ – Hydro-Québec)
• Patrick Lesage: Scientific developer (CRHQ – Hydro-Québec)
• Additional contributors from the Hydro-Québec research community may join as the project matures.

Sponsoring organization(s), along with any other key contributing individuals and/or organizations

Hydro-Québec (CRHQ)

Detail any existing community infrastructure, including:

• Github/GitLab, or other location where the code is hosted
• Website and/or docs
• Communication channels ( such as Mailing lists, Slack, IRC )
• Social Media Accounts
• PyPi, npm, or other App Stores maintainted by the project

No existing community infrastructur

Are there any specific infrastructure needs or requests outside of what is provided normally by LF Energy ? If so please detail them.

None at this stage.
Standard LFE GitHub, communication tools, and governance support are sufficient.

Why would this be a good candidate for inclusion in LF Energy?

The RAVEN targets a widely shared needs: accurate extraction and classification of distribution assets and vegetation from mobile LiDAR at utility scale. Hosting in LF Energy provides a vendor neutral home and governance, attracts contributors across utilities and vendors (including competitors).

How would this benefit from inclusion in LF Energy?

• Increased visibility across global utilities and research groups
• More robust code through external validation
• Collaboration opportunities on LiDAR standards, digital twins, and grid modeling
• Alignment with existing LFE projects working on digital grids (e.g., Grid Modeling, GridMetrics)

Provide a statement on alignment with the mission in the LF Energy charter.

RAVEN advances open, interoperable technologies that improve reliability distribution network reliability and resilience. By opening advanced LiDAR analytics to all utilities, it enables shared innovation and reduces duplicated R&D. It directly supports LF Energy’s mission to accelerate the global energy transition through collaborative work.

What specific need does this project address?

Utilities require reliable, automated solutions to map overhead distribution assets, detect power lines under occlusions and assess vegetation clearances. These capabilities are critical for digital twin development, operational safety, and vegetation management. The RAVEN project provides these functions with outputs compatible with GIS and enterprise asset management systems

Describe how this project impacts the energy industry.

RAVEN can lower inspection costs, reduce outages and safety risks through better predictive vegetation management and improve data quality feeding planning and maintenance. It helps utilities build accurate distribution level digital twins and supports condition based maintenance

Describe how this project intersects with other LF Energy projects/working groups/special interest groups.

RAVEN can serve as upstream data analytics building block for LF Energy projects addressing distribution network modeling, digital twins, and AI/ML workflows. The project will coordinate with relevant GISs/working groups to promote interoperability, common data models and shared datasets.
Potential Benefactors
• Electrical utilities
• Grid operators
• Forestry and vegetation management teams
• Research institutions working on LiDAR analytics and grid digitalization

Who are the potential benefactors of this project?

• Electrical utilities
• Grid operators
• Forestry and vegetation management teams
• Research institutions working on LiDAR analytics and grid digitalization

What other organizations in the world should be interested in this project?

National labs, universities in power systems and geospatial analytics, and companies building digital twins or vegetation management platforms such as:
• Jakarto
• Trifide
• Xeos
• AiDash
• Neara
• Skycraft
• Foresite

Plan for growing in maturity if accepted within LF Energy

In the next 18–24 months, RAVEN aims to:

1. Publish full documentation
2. Publish dataset (LiDAR point cloud sample)
3. Publish and upgrade the four algorithms
4. Develop a user community and update schedule

Project license

MPL-2.0

Is the project's code available now? If so provide a link to the code location.

Initial release will be published upon acceptance into the Sandbox.

Does this project have ongoing public (or private) technical meetings?

No public meetings yet.
Internal meetings at Hydro-Québec may evolve into open semestrial community calls.

Does this project's community venues have a code of conduct? If so, please provide a link to it?

Will adopt the LF Energy Code of Conduct

Describe the project's leadership team and decision-making process.

Currently driven by Hydro- Québec’s research team.
Initially led by Hydro-Québec researchers under a meritocratic, maintainer driven model. Goal: establish open governance with multiple organizations as maintainers and a documented decision process (lazy consensus, voting for contentious changes)

Does this project have public governance (more than just one organization)?

Not yet.

Does this project have a development schedule and/or release schedule?

This is our plan:
• April 2026: Documentation, dataset and algorithm for modeling obstructed power lines in mobile LiDAR
• October 2026: Documentation, dataset and algorithm for classifying power distribution assets in mobile LiDAR
• April 2027: Documentation, dataset and algorithm for span detection and extraction in mobile LiDAR data
• October 2027: Documentation, dataset and algorithm for detecting vegetation encroachment in mobile LiDAR

Does this project have dependencies on other open source projects? Which ones?

• Python scientific libraries (NumPy, SciPy, Pandas, GeoPandas)
• Point cloud processing tools (PDAL, Open3D, Lastool)
• ML frameworks (PyTorch or TensorFlow for NN module)

Describe the project's documentation.

Initial release will include:
• Project overview
• Algorithm descriptions
• Sample code

Describe any trademarks associated with the project.

No trademarks

Do you have a project roadmap? If so please attach or provide a link.

A detailed roadmap will be published on the GitHub project page.

Are this project's roadmap and meeting minutes public posted?

Not yet but planned once governance is established.

Does this project have a legal entity and/or registered trademarks?

No legal entity or trademarks currently.

Has this project been announced or promoted in any press?

Conferences publications are planned to enhance exposure.

Does this project compete with other open source projects or commercial products?

Commercial LiDAR analytics tools exist, but no open-source alternative provided

[richardklam]

Interested in this project and would like to be an advisor to the project, when it comes up for review. Some initial thoughts to maximize value across industry would be to define some technical minimal requirements as it comes to items such as lidar-point cloud resolutions, and key object recognition targets/uniform taxonomy/language.

[mohamedGaha1]

Thank you for your interest in the project.
Our algorithm uses as input a point cloud recorded from a vehicle‑mounted mobile mapping unit. The point cloud data is acquired using LiDAR sensors capable of generating up to 1 million points per second, combined with an inertial measurement unit (IMU). The resulting point‑cloud density typically ranges from 1,000 to 1,400 points/m², with a positional accuracy of approximately ±3 cm.
We fully agree that defining technical minimum requirements—such as LiDAR point density, spatial resolution, and a unified taxonomy for object classes—would help maximize interoperability and value across the industry.

Assessing algorithmic robustness under varying point densities is indeed an interesting and important direction. We have already conducted tests on our catenary‑curve detection algorithm under different density conditions, and the results were very promising. However, our other algorithms still need to be evaluated and analyzed under similar scenarios.
We appreciate your offer to act as an advisor during the project review phase and would be glad to continue the discussion as the work progresses.

[richardklam]

Just to add, I was referring to defining point cloud densities at the captured object say electrical pole, transformer, wire etc. which is a function of capture distance, and equipment capability. Considering that the capture platforms may be a vehicle, drone, or aviation asset, the density may be very low.