GitHub

By Osei.

I'll be the editor of Ziss.

Hard Krypton was informed that Hangzhou Run-off Technology Ltd. (hereinafter referred to as “Run-Technology”), a highly intelligent company in high-risk fields, had recently completed the first round of market-based financing (Pre-A round) with financial capital and commercial Tung Guangxi. This round will be used mainly for algorithm development, product matrix improvement and market expansion.

Founded in 2025 in Hangzhou, STD is a smart company that focuses on providing “robots+AI brain” solutions in high-risk industrial engineering areas (resource mines, power, oil and gas, traffic construction). The company ' s founder, Shu Jiang Peng, is a top-level national young talent, a researcher and director of the 100-person project at Zhejiang University. Since 2011, the team has been working on the development of smart urban Ai and engineering robotics.

Unlike the surveying robots that are common on the market, the core capability of truancy technology is not only the robot's movement control in complex engineering scenarios, but also the “engineer brain” that gives robots the ability to perform high-risk operations, especially “identify + diagnose + dispose” for the capital component.

"Most of the robots are now in the `scanter' phase, where they see problems, but they don't understand them, and they don't deal with them." Shukang Peng told the krypton. Through self-researched multi-modular models of engineering, the meso-sensor data, such as images, dots, ultrasound, electromagnetic waves, infrared, etc., can achieve millimetre-scale quantitative identification and state diagnosis of concrete cracks, steel structure erosion, rock and soil instability and hidden diseases, rather than simple qualitative diagnosis.

For the time being, it is mainly the light asset model of the “whole solution provider”. The company procures or works with the head manufacturer to produce the robotic body, with self-researched standardized “brain box” and uploading equipment, to produce a complete solution to the end customer. At the same time, some home-grown manufacturers lack vertical scene resolution capacity and may volunteer to push orders off and create synergies.

Another key advantage of the team is in the accumulation of data and downfall algorithms. Based on the background of the large earth and wood systems of the Zhejiang system, desert technology has the largest structured multi-modular database in the global infrastructure area (based on publicly available industry data) and has been cleaned and labelled, covering concrete/steel structures/soil surfaces and millions of specialized negative samples such as hidden diseases, equipment malfunctions, and human safety, with data volume levels ahead of peers by two orders and above. This creates significant barriers to their algorithms in terms of diagnostic accuracy and generalization.

Engineering applications digital twinning platform with multi-mechanical control and intelligent body analysis (graph source/enterprise)

With regard to commercialization, the company has opened the door. In the year in which it was established, lay-off technology contracted tens of millions of yuan with front-runner companies such as the Jiangxi copper industry, the national electricity grid and China ' s electricity construction. Products have been piloted in mining, tunnels, electricity, etc., to replace workers in high-risk transport.

“In the case of mines, for example, a high-risk work alley requires four workers in two shifts, and combined costs of carrying equipment exceed 1 million per year. And the maturity of a four-foot robotic product added to it costs hundreds of thousands of years after delivery, with a lifetime of up to three years. Not only are the cost advantages evident, but the risk of shutdown as a result of a safety accident is better avoided.”

According to Shu Jiang Peng, it is not yet possible to cover 100 per cent of all operations, and the remaining half still needs to be manual. Their goal is to increase coverage from 50 per cent to more than 90 per cent through data accumulation and iterative algorithms.

And Scramble Technology is very clear about its own future planning, "Doing the most engineering robot `brain', and using data and algorithms to solve the core problems of diagnosis and maintenance." Shukang Peng told krypton.

Sub-drives under the Four Foot Robotic Mines: millimetre-scale quantitative identification and diagnosis of mine rock bodies (source/enterprise)

Here is a selection of the dialogue between Krypton and Shujiang Peng:

Krypton: What are the technical advantages of hiking compared to other patrol robots on the market?

Shujiang Peng: Our technology core is high-precision and multimodular recognition. Most of AI does characterization, such as “there are cracks here”. We can identify how wide the cracks are, how far they go, to the millimetres and even sub-millimetres, which are essential for the maintenance of the project.

In addition, many of the deficiencies are hidden inside the structure and are not visible. We have integrated ultrasound, electromagnetic waves, infrared technology, so that we can see out of the inside of the structure. Behind this is the accumulation of our team in the capital+AI cross-section for more than 15 years, where our professional label negative sample size is a million-scale, while the industry generally has only a few thousand.

Scramble products: engineering brain KX and fully smart solutions (graph source/enterprise)

HKK: What are the biggest challenges that companies currently face in commercialized landings?

Shu Jiang Peng: It's mainly robotic stability and comprehensiveness. It's easy to run through a laboratory, but under a mine, a machine can be “stupid” when a network signal changes and an alley structure is fine-tuned. We are not yet able to cover 100 per cent of all operational sites, for example, by only half, with the remaining half still to be manual.

Our goal is to increase this coverage from 50 per cent to more than 90 per cent through the accumulation of data and iterative algorithms, and to allow clients to move from a “need-for-care machine” to a “can go home and sleep well”. But the trend is irreversible, because young people don't want to go down to the mines, you don't have to work without a robot.

Krypton: Do companies make human robots in future product planning? How do you view the four-foot and human-shaped course?

Shu Jiang Peng: In the short term, we still have four feet or four related products. The reason is simple, reliable. The field of high-risk industrial engineering should not be subject to error, and in the event of a large fall and loss, the stability of a complex terrain is far greater than it is now.

It may take three to five years for human robots to achieve the reliability of high-risk engineering scenarios. We will keep an eye on it, but it's not our focus, it's the field of our own manufacturers. We are well positioned — to do the most engineering robotic “brain”, to use data and algorithms to solve the core challenges of diagnosis and disposal.

Investor perspective:

Trader Tang Guoxiang investor: Specialty industries, high-risk security risks are serious and robotic logic is clear. As a result, the special robotics field is in great demand for capacity-building and has continued to expand. We look at the system programmes of emptied-land-mine cruises, inspections and even repairs that are available in a variety of settings, such as infrastructure, mines, transport, water, energy, electricity, chemical, etc., based on multi-modular models and smarts to form flexible and hard-to-hard transport inspection solutions. The team, with its technical background in Zheda, Stanford and Berkeley, and a well-integrated market-based team, has been very effective in achieving a settlement. Companies are expected to lead the special robotics industry, reshaping safe production patterns in traditional industries.

Financial capital investors: Financial capital has a long-term interest in the next generation of robotics with a “hand-brain-eye” synergy. The most central difference in hiatus technology is that it is not just a robotic body, but rather a moat with a multi-modular engineering model plus decades of industry-labelled databases, so that surveying robots and drones truly have the ability to “understand engineering defects and self-diagnosticate” and solves the industry pains of traditional motor robots.

The team, which originates from the Zheda-Stanford academic gene, has undertaken a number of national priority R & D programmes, has run through the closed loops of algorithms to orders in the real industrial scene, has worked with front-line customers such as Copper Gang Group, and commercialization has met expectations. As an investment agency in the province of Zhegen Zhejiang, capital capital will also help to accelerate product and market development, common service industry safety and digitization through the combined power of “capital + investment + industrial resources”.