Archetype Ai

0

How The 184-Year-Old Kajima Is Embracing an AI Future in Construction

Archetype AI Team
  • 3 min

In 1840, Iwakichi Kajima set up a small carpentry business in Edo, present-day Tokyo. Seeing the opportunity to revolutionize Japanese carpentry, construction, and building, Kajima built the first "Western-style" buildings in Yokohama. From there, the firm built railways, dams, power plants, and high-rises throughout Japan, Asia, and the world.

But even with this almost 200-year-old history, Kajima has never been a "traditional" company. What has forged their success is a progressive approach to technology adoption, always looking for ways to incorporate new technologies to streamline operations.

This approach has now led them to Physical AI and is why they started a partnership with Archetype AI — to bring Physical AI into their projects to reduce waste and improve efficiency for their construction management processes.

Managing Remote, Complex Construction Projects

The scale and complexity of Kajima's construction projects are immense. Thousands of workers, tens of millions of dollars of equipment, and hundreds of millions of dollars in materials mean every job is a massive organizational and engineering feat.

One such project is Kajima's five-year effort to widen an aging 100-year-old canal in Niigata, Japan. In the canal's estuary area, the cross-section is insufficient to safely discharge seasonal floodwaters, so improvements were necessary to reduce the risk of flooding to the residential area nearby. This area had originally been affected by flooding before the canal was built. However, after many years, it needed reconstruction.

A project like this presents huge challenges. It is a remote worksite spread over a vast area. Project managers need daily insight into the work, progress, and safety on site, even when they can't monitor it in person. For example, if a critical piece of equipment breaks down, they must quickly reallocate resources and revise the schedule; or if weather conditions suddenly change, the work plan must be adjusted to ensure worker safety.

"Field personnel face strict demands from clients to thoroughly oversee site operations. Meanwhile, the construction industry is experiencing a decline in workforce, resulting in an increasing workload per employee," explains Shinnosuke Sekihara, Mechanical and Electric Deputy Manager at Kajima. "In cases where process control fails and completion is delayed, clients may demand millions of yen for each day of delay."

Understanding the jobsite so that the necessary action can be taken at the right time is one of the key factors for project success. The inability to do so efficiently contributes to significant cost overruns in large-scale construction projects. Research from McKinsey reveals that 98% of construction megaprojects exceed their budgets, with average cost increases of 80% above the original value and schedule slippage of 20 months.

To address inefficiency that significantly impacts project costs across the industry, Kajima aims to transform both their own operations and the industry at large by leveraging cutting-edge technologies to achieve 50% remote management of job sites. This ambitious vision is becoming possible with Physical AI.

Turning Terabytes of Data into Insights

Archetype AI has collaborated with Kajima to develop a Physical AI pilot that meets their unique needs. We worked with terabytes of historic video from 27 cameras over four years to demonstrate how Newton can turn this data into actionable insights that help the Kajima team avoid cost overruns and improve overall efficiency. This vast dataset comes from a camera network spread across the massive construction site, incorporating information from every stage of the multiyear project. We then augmented this video dataset with years of site-specific time-series weather data to provide additional context to this visual data.

Newton must contend with the real-world video quality challenges of a construction site: extreme weather, equipment movement, obstructed views, changing seasons, and fluctuating river levels. "It was impossible for humans to find the right video of the targeted drilling operations from 12,000 videos; Newton was able to extract these targeted videos seamlessly in seconds," says Shinnosuke Sekihara, highlighting the scale of the data challenge.

Here is how Archetype AI helped Kajima tackle these challenges:

  • Newton Semantic Lenses: These interpret the vast amount of video and weather data in several key ways, allowing project managers to focus on specific aspects of the construction process.

A Daily Log Lens visualizes start and end times for different work periods and machine operations, giving managers a detailed view of daily activities and helping them find inefficiencies and understand why productivity varies depending on weather conditions.

A Summary Lens offers an aggregate view of the work done over specific date ranges, allowing for broader trend analysis and progress tracking.

  • Video playback with object detection: Project managers can view specific video segments with the AI-detected objects highlighted, providing visual verification of the AI's interpretations for the user.

Archetype AI's Newton model is designed to learn and evolve based on user interaction. As Kajima's team uses the tool and labels the causes of work deviations, the Lens starts learning from these inputs. This enables the AI to propose explanations when it detects factors that could cause deviations, such as high tides or other environmental conditions.

"Initially, we were considering asking questions about the videos in a chat format, but due to the wide variety of questions and the ambiguity involved, we realized that having a specialized Lens would be more helpful," shares Shinnosuke Sekihara. Over time, these Lenses have become increasingly tailored to Kajima's specific needs and challenges, providing ever more valuable insights and predictive capabilities to support efficient project management. "In the future, we envision using Newton to analyze not only what work was done in the past but to monitor construction sites in real time, highlighting different factors that affect productivity and contribute to success or failure. That way, Newton can help remote project managers react quickly and make better decisions."

Archetype AI and the Future of Construction Industry

With the Kajima pilot success, Archetype is continuing to work on solutions for the construction industry. "In the future, we expect that the AI will accumulate feedback from field experts, further expanding its application scope," says Brandon Barbello, co-founder and COO of Archetype AI. "Newton can serve as an important interface for preserving know-how as skilled engineers retire—an ongoing industry challenge."

Here is how Archetype AI's platform can serve construction companies in the future:

  1. Improved project planning and execution. By drawing insights from past work, project managers can manage similar projects more efficiently. We foresee experts using Newton for monitoring excavations in real time to take measures against varying weather conditions and improve work plans as the project unfolds.
  2. Enhanced accountability for subcontractors: In the future, project managers will be able to get detailed activity logs and optimize subcontractor management.
  3. Preservation of expert knowledge: Newton captures and preserves expertise from veteran construction professionals, ensuring critical insights aren't lost as experienced staff retire.

We strongly believe that Newton's capabilities can be applied across the entire construction sector. Its use cases include, but are not limited to, worker safety monitoring, planning scenario modeling, and better coordination of equipment, vehicles, and material deliveries — all important factors in reducing the industry's cost overruns and delays.

By turning the vast amounts of sensor data from construction sites into actionable insights, Newton helps experts make better decisions faster, reduce waste, improve safety, and deliver projects on time. As companies face increasing pressure to modernize and improve efficiency, Newton can turn construction's biggest challenges into opportunities for improvement and growth.

Recommended posts

This was the year Physical AI moved from concept to reality, as the importance of using modern AI foundation models to solve real-world problems entered the mainstream. We are proud to be at the forefront of this movement. Let’s look at the year’s most defining moments from Archetype AI.

Imagine a world where your car responds not just to what you see but to what every vehicle, traffic light, and smartphone detects. Soarchain is making this a reality by combining their decentralized platform with Archetype AI’s developer API, Newton. Together, we will give drivers and autonomous vehicles a rich real time understanding of what’s happening on the roads around them.

August 13, 2024

Implementing AI in industrial settings comes with significant challenges like ensuring employee safety, estimating productivity, and monitoring hazards—all requiring real-time processing. However, sending sensor data to the cloud for analysis introduces latency and security concerns, driving up costs. The solution? Eliminate the cloud. With Archetype AI’s Newton foundation model, AI can run on local machines using a single off-the-shelf GPU, delivering low latency, high security, and reduced costs in environments like manufacturing, logistics, transportation, and construction.

September 5, 2024

At Archetype, we want to use AI to solve real world problems by empowering organizations to build for their own use cases. We aren’t building verticalized solutions –instead, we want to give engineers, developers, and companies the AI tools and platform they need to create their own solutions in the physical world.

In a world where devices constantly collect and transmit data, many organizations struggle to harness its potential. At Archetype AI, we believe AI is the key to transforming this raw data into actionable insights. Join us as we explore how the convergence of sensors and AI can help us better understand the world around us.

August 26, 2024