Muqing Li: Pioneering Robotic Vision for Autonomous Solar Deployment in the Age of AI and Renewable Energy

ARLINGTON, Va., June 30, 2025 /PRNewswire/ -- Muqing Li has been assigned as the Lead Performance Engineer for the largest solar-plus-storage project in the United States, commissioned by AES Corporation, a Fortune 500 global energy leader. At the forefront of digital transformation in the renewable energy sector, Mr. Li is driving a new era of intelligent deployment powered by breakthroughs in solar robotics.

His most recent innovation, an advanced SLAM (Simultaneous Localization and Mapping) system, integrates state-of-the-art computer vision and route planning to enable autonomous operation in complex, unmapped environments. For the first time in the industry, this technology was embedded into a production-level solar installation robot, fusing computer vision and motion control into a fully functional algorithm for autonomous deployment. This advancement enables the rapid and scalable deployment of renewable energy into critical markets, directly supporting the surging energy needs driven by AI and data infrastructure, and reinforcing Mr. Li's pivotal role in transforming solar technology.

As all tech-whales desperately demanding clean and reliable source of energy to evolve the AI technology, Mr. Li's work is setting the foundation to intelligently meet the growing energy demand. His groundbreaking work has been particularly impactful in SoCal, facilitating the construction and integration of 2GW into the CAISO energy market since 2023. This robotic technology has already been deployed on the largest solar-plus-storage site in the United States, Bellefield, 2GW renewable power plant in California, which accounts for nearly 10% of the entire state's solar capacity. Effectively, this 2GW project break the scale and speed of construction, and will continue to generate equivalently 467,000 homes of electricity annually, reducing over 1 million metric tons of CO₂ annually. Mr. Li's pioneering work in computer vision (CV) and real-time localization is laying the groundwork for the next generation of autonomous solar construction technologies.

Mr. Li's localization and mapping system represents a core innovation in robotic perception. It combines stereo camera inputs with inertial motion data through a mathematically rigorous Extended Kalman Filter (EKF), allowing robots to sense and map unstructured environments with human-like awareness. The system incorporates over 100 custom mathematical models in sensor fusion. The mathematical model combines vision camera, a variety of depth sensors, lidar scans to emulate intelligent perception with a distance-accuracy level of 0.25cm. Ultimately, it enables the robot to understand and adapt to complex field conditions and brings human-level intelligence to machine navigation.

Released under the MIT license, Mr. Li's SLAM system reflects both technical sophistication and a commitment to scientific transparency. This SLAM code library has gained global recognition, with GitHub stars and forks from engineers and researchers at top institutions such as Purdue University, Vanderbilt, and NTNU. The platform is now viewed as a benchmark in vision-aided robotic localization, and its adoption underscores its value as a critical enabler of autonomous motion in real-world industrial settings.

Alongside his SLAM development, Mr. Li co-authored the landmark paper "Research on Image Classification and Semantic Segmentation Model Based on Convolutional Neural Network," which introduced EDNET, a CNN architecture for high-precision object detection and semantic segmentation. The model achieved nearly 11.7% improvement in mIoU over traditional networks, demonstrating strong applicability to boundary-aware image recognition. This EDNET technology directly supports the solar robot's bracket recognition engine, allowing it to detect and adapt to various OEM solar panel designs. This enhances installation accuracy, reduces safety risks from module misalignment, broadens the range of solar panels that can be reliably deployed, and ultimately supports grid resilience and plant safety.

What distinguishes Mr. Li is his rare combination of scientific rigor, open-source leadership, and direct industrial impact. His SLAM codebase is not just a research artifact; it is already embedded in billion-dollar renewable energy infrastructure. His deep learning models are not limited to laboratory benchmarks; they drive real-time decisions by intelligent machines operating in solar fields across the United States.

Mr. Li's expertise in AI and automation is reshaping how solar infrastructure is constructed and maintained. At the convergence of robotic vision, machine learning, and renewable energy automation, he plays an essential role in automating performance diagnostics and intelligent operation diagnostics across utility-scale renewable energy projects. His contributions extends beyond robotics to encompass statistical benchmarking systems to track asset-level performance during grid disturbances, supporting compliance and resilience under CAISO protocols. Through his technical leadership, Muqing has helped transition 3 GW of clean energy into commercial operation and in the pipeline. His work on solar robotics marks a transformative step toward automated and scalable clean energy deployment.

As solar robotics moves from experimentation to standard practice, Muqing Li's originality continues to drive the frontier of renewable energy automation. From robust perception systems to scalable mapping pipelines, his innovations exemplify how computer vision and robotics can accelerate the global clean energy transition and advancement of AI, at a scale of safety, precision, autonomy.

For more on AES Robotic: https://www.aes.com/maximo

For more on Bellefield: https://www.aes.com/california/project/bellefield-solar-storage

Project Status: AES completes first phase of the largest solar-plus-storage project in the U.S.

Media contact: Steven He, [email protected]