Vehicle-mounted cameras are a strong contender for transforming the wautility asset management is done. Read on to learn why!
So far in this series of articles on AI technologies innovation leaders must consider for utility asset management, we have shared in-depth insights into drones and satellites.
For this article, we will be shifting gears to focus on another emerging AI-powered technology: vehicle-mounted cameras. When implemented correctly, it’s a must-consider for innovation leaders looking to adopt new grid management and assessment strategies.
Much like the format of the previous articles, we will use the same consideration framework by discussing: data type and quality, cost, ease of use, scalability and safety.
Within the realm of utility asset management, the importance of data cannot be overstated. Vehicle-mounted camera systems have emerged as powerful tools in this pursuit, serving as a versatile platform that can accommodate a variety of different sensors beyond that of conventional RGB imagery.
First and foremost, one of the major benefits of vehicle-mounted systems is their proximity to the poles and assets they inspect. As these vehicles travel down roads, they inherently gather higher fidelity data than those of a remote solution such as satellites.
While high-resolution cameras can capture poles and pole-top components in detail, allowing for detection of asset tags, transformer ratings, and more, the incorporation of other sensors greatly expands the utility of these types of systems. Infrared cameras can detect equipment hotspots that may lead to outage. LiDAR enables highly accurate measurements by generating point cloud data. Another technique that is increasing in popularity is the use of stereovision. Stereovision is a technique used in computer vision and robotics that mimics the human visual system's ability to perceive depth and three-dimensional information from two slightly offset images captured by two cameras, just like our two eyes. Although stereovision is not always as accurate as LiDAR, it is significantly more cost effective and less sensitive to weather and lighting conditions. With this cheaper method of capturing depth information, use cases that require the ability to make measurements (think: pole loading analysis, vegetation encroachment, NESC clearance measurements, the list goes on) can be automated.
In summary, vehicle-mounted systems provide not only granularity, but a variety of data types that can translate into actionable insights. This data-rich approach can foster heightened situational awareness, allowing utility leaders to address issues at their root and proactively ensure a resilient network.
When considering vehicle-mounted camera solutions, there are two primary options: outsourcing to a third party or procuring/developing the technology in-house. Similar to the drone scenario, each approach carries its own pros and cons.
The primary option within this space is the as-a-service (aaS) model, or the "black box" approach. Companies that offer this option typically come with their own vehicles, proprietary sensor systems and specialized operators. Although this option provides convenience, running these data collection circuits on something as widespread as the distribution network can quickly rack up costs (in particular, O&M costs). Additionally, once the data collection process has been completed, utilities still have to determine the best way to process the data as the service provider may not specialize in data analysis. This may incur additional O&M costs depending on the in-house capabilities of the given utility.
On the flip side, there's the option to invest in developing their own camera systems. Procuring hardware can be counted as capital expenditures (CapEx), which help the utilities’ bottom line.
Although there are accounting advantages to procuring in-house camera hardware, knowing which systems to acquire can be a daunting task. While sophisticated survey-grade camera systems such as the Ladybug and Mosaic X exist and can enable detailed data capture including high-resolution imagery, they do not come with the requisite data analytics capabilities required to transform captured data into actionable insights. Simply put, using an off-the-shelf system to collect grid data without an accompanying analytics is an integral part, but only one part, of the solution..
The current state of vehicle-mounted camera systems is fragmented, with varied levels of usability across different service providers and utilities. As we’ve been discussing, there are two parts that need to be solved in order to have a successful utility management solution: data collection and data analysis. More often than not, parties have solved for one or the other.
Third-party service providers offer specialized camera systems for data collection; however, as mentioned in the previous section, these providers typically have limited analysis capabilities as they are industry agnostic—that is, electric utilities are just one customer category. As a result, utilities then need to either develop in-house solutions or go to yet another third party in order to extract insights from their data. Not to mention that there may also be limitations as to how the data can be utilized given that it was collected via a proprietary system.
As an alternative, some utilities have opted to procure their own off-the-shelf 3D mapping and survey cameras, which grants them full ownership of their data. However, the same challenge still applies: in order to fully leverage the collected data, there needs to be supporting analytics. For example, will imagery and point cloud data need to be combined to extract pole geolocations? Are there defect detection models that need to be built in order to extract potential outage-causing issues?
Depending on the utility, there may or may not be dedicated data teams or existing in-house solutions capable of addressing these needs. In short, to truly have a simple, easy-to-use system, utilities need to consider the entire workflow starting from data collection and ending with actionable insights that directly address challenges in utility management.
The key challenge with finding a scalable utility asset management solution today lies in finding cost-effective technologies that can deliver high-quality data covering large quantities of assets. Drones offer detailed imagery but their coverage is hampered by FAA and BVLOS regulations. Satellites provide broad coverage but lack the imagery resolution for in-depth analysis.
This is where vehicle-mounted cameras step in as a practical solution—if implemented correctly.
Unlike third-party service providers, which can quickly accumulate O&M costs, integrating cameras into existing utility fleet vehicles can be categorized as capital expenses, which help reduce costs significantly. The key shift lies in reimagining asset management workflows—from a meticulous examination of 10% of assets to a broader, less thorough assessment of the 80% of assets. Vehicle-mounted cameras, combined with AI technology, can help make this a reality.
Utilities already have fleets of vehicles. Equipping these trucks with cameras offers significant coverage, especially as the majority of distribution assets are accessible roadside. AI comes onto the scene as a cost-effective, first-pass reviewer. Trained to detect high-priority defects, AI models can identify potential outage-causing issues during routine utility operations, reducing the need for dedicated inspection truck rolls. This synergy between vehicle-mounted cameras and AI not only enhances grid situational awareness but also ensures a proactive approach to maintaining utility assets without exhausting resources on meticulous inspections.
Implementing a vehicle-mounted camera system for utility asset management offers tangible safety benefits. By reducing the need for in-person visits, the system can help reduce the risks associated with on-site inspections. Additionally, by leveraging machine learning models that can identify certain potential defects from afar, utilities can proactively perform grid maintenance and hardening work. This approach not only ensures field workers' safety but also enhances overall operational safety by preventing accidents and failures beforehand.
Noteworthy’s Inspect solution stands out as a cost-effective and scalable end-to-end solution specifically designed to meet the requirements of utility operations. Noteworthy Inspect easily installs onto existing fleet vehicles, autonomously collects high-quality grid data in a systematic fashion, and analyzes collected data to extract meaningful insights that translate to direct work plans on the grid. With Noteworthy’s solution, utilities can vastly improve their grid situational awareness and make better strategic decisions with respect to increasing grid reliability, resiliency and safety.
Noteworthy’s system addresses the entire workflow from data collection to analysis, minimizing friction between steps.
Our solution operates completely autonomously—no training or specialized operators required. Once our cameras are mounted, they seamlessly activate and deactivate in sync with the vehicle ignition, eliminating the need for constant human intervention. We offer several different data extraction methods tailored to each customer's unique needs, ensuring that the collected data is accessible.
During the data collection process, Noteworthy’s on-edge machine learning models go through a preliminary filtering process of the imagery and metadata. As the vehicles drive, our system detects assets of interest (in this case, utility poles), geolocates them and captures high-resolution imagery of these assets.
However, we know that in order to derive true value, utilities need more than just collections of images along lat/lon coordinates. At Noteworthy, we leverage our consistent data collection method to build models that help utilities derive actionable insights from their data. Our models identify and inventory pole-top components, potential outage-causing defects, third party attachments, and more.
In summary, Noteworthy's solution offers a seamless end-to-end workflow for utility asset management. Integrated into existing fleet vehicles, it autonomously collects high-quality grid data while our advanced machine learning models provide actionable insights, empowering utilities to make more informed decisions that benefit the grid and subsequently, their ratepayers.
Pros:
Cons:
