This site is fictional demo content. It is not real news or affiliated with any real organization. Do not treat it as fact or professional advice.

Full article

FULL TEXT

View this issue
Deep diveROBOTICS

High-Voltage Grid Autonomous Maintenance Robot GridSweep: Walking on Live Lines to Eliminate Defects

GridSweep, jointly released by the China Electric Power Research Institute and the State Grid Corporation, is a robot that can walk autonomously on 1,000 kV ultra-high-voltage live lines and complete maintenance. In a trial run on the East China Grid, the robot identified 780 line defects, of which 120 were repaired in time, avoiding potential large-scale blackouts.

GridSweep, jointly released by the China Electric Power Research Institute and the State Grid Corporation, is a robot that can walk autonomously on 1,000 kV ultra-high-voltage live lines and complete maintenance. The robot uses a specially designed bipedal structure to attach to high-voltage conductors and can autonomously walk for hundreds of kilometers, performing line defect detection and small-scale repairs while the line remains energized.

The project grew out of the extreme danger of high-voltage grid maintenance. Ultra-high-voltage transmission lines reach one million volts, and lines are typically more than 50 meters above the ground. Traditional manual inspection requires live-line operations or power-off maintenance. The former is extremely risky, and the latter has a wide impact range. The CEPRI team spent six years on research and development, eventually building a maintenance robot that can work autonomously in live-line conditions.

The core technology is a bipedal adsorption walking mechanism. Each foot of GridSweep is equipped with a hybrid adsorption system combining permanent magnets and electromagnets, which can stand firmly on a live conductor. When walking, the electromagnet is energized to generate a repulsive force that lifts the foot off the conductor, and the mechanical structure takes a step forward. When the electromagnet is de-energized, the permanent magnet re-attaches the foot to the conductor. The entire walking process does not require cutting off the current and has zero impact on grid operation.

For the perception system, GridSweep carries a high-definition camera, an infrared thermal imager, an ultrasonic sensor, and a UV imager, capable of simultaneously identifying multiple types of faults: visible defects (breaks, wear), temperature anomalies (poor contact), and internal damage (corona discharge). All sensor data is transmitted back to the control center in real time via 5G. The control center's AI system classifies defect severity and decides whether the robot will repair on site or whether personnel will be dispatched.

For deployment results, State Grid conducted a trial run with 24 GridSweep robots on the East China Grid. During the 12-month operation, the robots autonomously walked a total of about 86,000 kilometers (equivalent to more than two laps around the Earth), identifying 780 line defects. Of these, 120 were small-scale defects (such as a single loose screw or local wear) that the robot repaired on site, avoiding 12 potential large-scale blackouts. Zhou Xiaohang, who heads equipment at State Grid, said the use of GridSweep has improved grid inspection efficiency by about 15 times while avoiding high-altitude live-line operations for maintenance personnel.

On the commercialization path, GridSweep is custom-procured by State Grid at about 1.2 million renminbi per unit, mainly for the operation and maintenance of domestic UHV lines. State Grid plans to deploy 200 GridSweep robots by 2027, covering the entire UHV backbone grid. CEPRI is in talks with power companies in Brazil, Russia, and South Africa about export versions, adapted to different voltage levels in those countries (Brazil 800 kV, South Africa 765 kV, Russia 1,150 kV).

Critics point out that the AI defect recognition system of GridSweep may still miss rare fault types. CEPRI responded that the team has built a training database containing 120,000 defect samples and updates it monthly. At the same time, the robot retains a manual remote-control mode, allowing experienced operations staff to take over in complex situations. This hybrid mode combining AI autonomy with human fallback is currently the optimal approach for grid robots.