Neha Joshi’s introduction to ANYbotics came in 2024, when a colleague deployed one of the company’s quadruped robots at a power facility in Israel. Within weeks, that initial exposure evolved into a cross-continental workflow—processing inspection data in Schenectady and validating insights through a new deployment in Ireland.

The robot demonstrated strong field performance: navigating stairs, accessing confined areas, capturing high-resolution imagery, conducting thermal analysis, and identifying acoustic anomalies. However, a critical bottleneck emerged—not in data collection, but in data utilization.

According to Joshi, product leader for AI and machine learning at GE Vernova, large volumes of inspection data remained stored locally on-site, limiting accessibility and preventing advanced analytics. This constraint highlighted a broader industry issue: valuable operational data often remains underexploited due to infrastructure limitations.

Integrating Robotics, AI, and Cloud Intelligence

This realization became the catalyst for collaboration between GE Vernova and ANYbotics. The joint initiative focuses on combining autonomous robotics with artificial intelligence and cloud computing to modernize industrial inspection workflows.

As a major OEM supporting roughly a quarter of global energy generation, GE Vernova is prioritizing technologies that enhance operational safety and efficiency. The integration of robotics into its Asset Performance Management (APM) ecosystem reflects a strategic move toward predictive, data-driven maintenance.

Autonomous robots now serve as physical extensions of AI—capable of gathering data in environments that are hazardous or inaccessible for human inspectors. These environments often involve high voltage, extreme temperatures, or confined industrial spaces, making robotic inspection both safer and more scalable.

From Data Overload to Actionable Insights

The impact of this integration is significant. In one case, a dataset of 400 inspection images—covering components such as flanges and pipelines—would traditionally require two weeks of manual analysis. Using computer vision models, the same dataset was processed in just 30 minutes.

The ANYmal robot, developed by ANYbotics, functions as a multimodal data acquisition platform. It combines LiDAR, RGB imaging, ultrasonic sensing, and gas and thermal detection, producing complex datasets that exceed the capabilities of traditional on-premise systems.

By transferring this data to the cloud, GE Vernova’s APM platform converts raw inputs into structured time-series data. This enables operators to detect trends, anticipate failures, and integrate insights with existing asset knowledge.

Initial results from a proof of concept in Israel indicated measurable improvements, including:

• Reduced unplanned downtime
• Faster and more accurate fault diagnostics
• Up to 40% reduction in reactive maintenance and HSE-related costs
• Validating Autonomous Inspection in Real Environments

To verify system performance, a two-week pilot was conducted at GE Vernova Advanced Research Center in Schenectady. The first phase focused on testing the robot across diverse inspection scenarios, including leak detection, corrosion analysis, and acoustic monitoring.

During the second phase, collected data was processed through the Autonomous Inspection application—an AI-driven module within the APM ecosystem. The results aligned closely with traditional manual inspections, confirming the reliability of the approach.

Following this validation, the system was deployed at a customer site in Ireland. The implementation roadmap includes staged capabilities, starting with thermal monitoring and expanding to ultrasonic diagnostics and gas detection.

Scalable Applications Across Energy Infrastructure

While initial deployments focus on gas turbines, the architecture is both robot-agnostic and asset-agnostic. This enables application across a wide range of industrial environments, including:

• Power transformers
• Chemical processing facilities
• Oil and gas infrastructure
• Data centers

The long-term vision extends beyond passive inspection. Future iterations may allow APM systems to autonomously direct robots—assigning follow-up inspections based on detected anomalies. For example, if corrosion is identified, the system could schedule a targeted reinspection without human intervention.

Human Oversight Remains Central

Despite increasing automation, decision-making authority remains with human operators. Robots and AI enhance situational awareness and accelerate diagnostics, but strategic and safety-critical decisions continue to rely on human expertise.

The collaboration between GE Vernova and ANYbotics demonstrates how combining robotics, AI, and cloud infrastructure can fundamentally transform asset inspection. The result is a more proactive, efficient, and safer operational model—aligned with the evolving demands of the global energy sector.

Source