Blockchain-Driven iPaaS for Revolutionizing Construction Management with ACC

FUMIYA MATSUSHITA: Hello, everyone. Welcome. Very welcome to our session. I'm very honored to have this session. And thank you very much to Autodesk team to give me this opportunity. So today I would like to give you a presentation about blockchain-driven iPaaS for revolutionizing construction management with ACC.

First, let me share one keyword, R-CDE. R-CDE represent for Reliable Common Data Environment. Please remember my session title is Blockchain-Driven iPaas. The name of iPaaS, which we develop, is R-CDE. Most of the participants here have already understood the importance of CDE. Maybe half of participants don't know that. The importance of CDE is widely recognized because everyone here know the 365. Nowadays, we call it Autodesk Construction Cloud, ACC. We can utilize this great service in project.

Then that is why-- I think some of you may have this question. Why you, your team develop CDE? Also, what is our level? Please remember, again, my title is Blockchain-Driven iPaaS. The reason for its high reliability is that it use blockchain. But still you may have the question why high reliability is needed.

Let's start the presentation with this question in mind. This project has been developed with many people and a lot of time. Today, I would like to share a part of our long journey from us. Before jumping into the content, let me introduce us briefly.

I'm Fumiya Matsushita. I work for Shimizu Corporation as civil engineer and also work in University of Tokyo as project assistant professor. And my company, Shimizu Corporation is one of the leading company in Japan and launched more than 250 years ago.

So 90 minutes is quite short time to introduce my company. So that is why if you are interested in my company, please Google it. So let me introduce my colleague, Masamiki. So please go ahead, Masamiki. OK.

MASAMIKI MATSUBARA: I'm Matsubara. So I'm the CTO of AMDlab. My career path is unique in that I started as an architectural designer before transiting to become a system engineer. Next, [JAPANESE]. Let me tell you a bit about AMDlab. Despite having AMD in our name, we have absolutely no connection to the famous CPU and GPU company. The name AMD simply come from the initials of our founder.

[SPEAKING JAPANESE]

We are a company that has grown by developing beam plugging and automatic design algorithms. For instance, we've developed algorithms that optimize parking lot layouts just by drawing lines.

[SPEAKING JAPANESE]

--or algorithms that automatically design restroom. We've done a lot of work in automation. This automatic restroom design service was developed in collaboration with Richard Corporation. And I actually spoke about it at Autodesk University 2021. For those interested in this topic, please check out the archived video from 2021. That concludes my brief introduction. OK, thank you, [JAPANESE].

FUMIYA MATSUSHITA: OK, so let's go back to the content. Our session consists of mainly three parts, first, project background and overall picture from me. Second is a system for data-driven approach from Masamiki. And the last part is open ecosystem for social implementation.

OK, so first part, I'm going to speak about project background and overall picture as overview. Let's begin with social background. While the global issue is labor shortages, Japan is also no exception. And serious labor shortage is predicted as a result of continuing population decline.

Even with a neutral perspective, a population decline of approximately 20% is expected over the next 20 years from 2020 to 2040. This problem is a very serious problem in terms of the lack of engineers and workers in the construction industry.

And also, aging infrastructure, major problem for our industry-- in Japan, the percentage of infrastructure over 50 years old is increasing. This figure shows the percentage of over 50 years old. Let's see below the grid, the left one. The percentage of over 50 years will reach more than 70% in 2040. It mean we need to renew the infrastructure.

In short, we have three points, three major points to consider. Because of aging of infrastructure, we need to deliver the infrastructure. And adding to this point, we also consider the natural disaster because recently natural disasters have been intensifying due to climate change and other factors.

It means we need-- civil engineers need to tackle this problem. And because of these two point, aging of infrastructure, natural disaster, the demand for civil engineer and worker is increasing. But the point is labor shortages. So we need to improve our productivity.

So as a conclusion, as a key challenge, we need to improve the productivity. But we have good news. In Japan, the implementation of ICT in public work is increasing. This implementation of ICT is conducted under the policy of i-Construction, which launched by Ministry of Land Infrastructure, Transport, and Tourism, MLIT, means government.

So based on the statistics, recent implementation rate is 85%. This is extremely rare even when you look around the world. And I think it is a strength of the Japanese construction industry. So you may not familiar with ICT work. So let me introduce some points.

So ICT work implement ICT technology to the construction site. So, for example, when we construct a road, we need to compact the soil. So in this case, we need to manage how many number are we? We need to manage the number of the compaction.

But, in this case, we can utilize the GNSS. This case is a [INAUDIBLE] one. And, also, as build inspection, we can utilize the point cloud. And regarding the excavation, recently we implement semiautonomous excavator to the site. So thanks to this effort, we can gather, I mean, the data being collected relatively easy.

But the point is original workflow still remains. Even we get the data from the site, we share the data through the email or chat, so, I mean, original method. Everyone knows it is time consuming. In order to efficiently share this-- excuse me, in order to share efficiently this data among the players involved in the construction production process and realize the digital workflow, a common platform specialized for ICT construction is required.

As I said, Japan's ICT construction implementation rate is quite high. That is why this common data platform is necessary. And we develop this platform with utilizing ACC. We call this common data platform as R-CDE.

But still you have this question. Why blockchain is necessary? So, as an example, I introduced management of the number of loading compaction utilizing GNSS. But the point is, this data is produced by contractor.

Of course, we, contractor, have never, ever tampered the data. But the point is that there is a possibility of the tampering. That is why, now, in Japanese regulation, on-site inspection by client is required. Beginning of our project, we conduct one survey to understand the working ratio of contractor from more than 3,000 question area.

And as a result, we found this inspection work accounts for 20% of our working relation. That is why, if we can prove the authenticity of the data between client and contractors, we can achieve a 20% efficiency improvement in our industry. This is very big. This is a huge impact to our industry.

Let me show how to utilize the blockchain in our industry. During the construction phase, we can gather a lot of information, for example, the bucket age represented for the [INAUDIBLE] build. And, also, surveying information can utilize two major progress.

So these data from these instrument can directly input to the blockchain system through the API. The inspection on the payment states to release the data from the blockchain system and contractor contract conduct quality and attribute inspection. Regarding the survey information, we can utilize for progress measurement.

And, also, we developed one smart contract program so we can input contractual information. And based on this contractual information, we input contract performance information based on the inspection. And utilizing this data, we can connect with this data to the payment. And we can realize semi-autonomous payment as well.

And, also, as a point, we need to solve-- to realize this workflow, we need seamless data integration. That is why we developed the API for our system. This point is explained from Masamiki. So please go ahead, Masamiki, for your part.

MASAMIKI MATSUBARA: All right, now, go ahead and talk about the system. Next. First, I talk about workflow.

[SPEAKING JAPANESE]

It's important to explain that there are two types of users with different roles in R-CDE. One is author. An the other is contractor. The workflow cannot progress with just an author or just a contractor. This application is designed for author and contractor to communicate and advance the workflow together.

[SPEAKING JAPANESE]

Real client and contractor use this application for construction project. The client becomes the orderer.

[SPEAKING JAPANESE]

And the contractor remains the contractor.

[SPEAKING JAPANESE]

When your contractor and subcontractor use this application, the contractor becomes the orderer.

[SPEAKING JAPANESE]

And the subcontractor becomes contractor to progress through this workflow.

[SPEAKING JAPANESE]

The system is designed to accommodate their multi-layered subcontracting structure typical in construction sites. Client can invite contractor to the project. Contractors can invite subcontractor. And subcontractor can invite sub subcontractor, blah, blah, blah. Each [INAUDIBLE] is an auditor or contractor, according to their respective contract, advancing the workflow.

[SPEAKING JAPANESE]

The workflow begin with the orderer creating construction site. They then create a contract and invite contractor to fulfill the contract.

[SPEAKING JAPANESE]

This is the site creation screen. Enter the site name, addresses, contract date, and then completion date, contract amount, and advance payment rate to create the site. After creating the site, you create a contract for that site.

Contracts can be created as either an order or contractor. Here, you enter the contract information unit, price, and quantity to create the contract. Subsequently, an invitation email is sent to the contractor.

[SPEAKING JAPANESE]

The invited contractor after completing their contracted work capture point cloud data of the site and upload it, R-CDE.

[SPEAKING JAPANESE]

The type of file that can be uploaded are determined based on two use cases, one for Earthwork. It's expected to upload point cloud and design data. For NATM tunnel construction, design data alignment information and point cloud as construction management information are expected to be uploaded.

[SPEAKING JAPANESE]

For Earthwork, first, upload the design data, like this, and then after that, the Point Lookout data. So we have to enter some information for the point cloud. After that, so we verified the data in the viewer.

[SPEAKING JAPANESE]

Similarly, for NATM tunnel, upload the tunnel design data. In this case, so I'm using an IFC file for that, like this, and then upload alignment information.

So now we can see a line of the thumbnail. And then we upload point cloud. So it's also need some metadata for the point cloud. So we check for any issues in the data on the build.

[SPEAKING JAPANESE]

The next step is for the contractor to select area for evaluation.

[SPEAKING JAPANESE]

The contractor place anchor points on the viewer to define the area needed for inspection. So if you click, we can see the anchor point area. And then they specify the point cloud and anchor point, perform point cloud trimming, like this.

[SPEAKING JAPANESE]

The next is evaluation by the contractor.

[SPEAKING JAPANESE]

One of R-CDE's crucial feature is generating a heatmap by calculating the height difference between the point cloud and design data. This heatmap is used to identify area where construction may not have been carried out according to the design. So we can create heatmap like this.

[SPEAKING JAPANESE]

After confirming there are no issues, an inspection request is made.

[SPEAKING JAPANESE]

The contractor specifies the other build data, like heatmap, showing the completion of contract work, along with the source point cloud and design data, so now are using our files, and then input some data for inspection. Yeah. So now I send inspection to the orderer. As R-CDE functions, R-CDE file submitted for inspection change the state to shared.

[SPEAKING JAPANESE]

After confirming there are no issues on inspection-- sorry. So the final step in the workflow is the order inspection.

[SPEAKING JAPANESE]

The orderer accesses the inspection request screen, reviews the data, and other-- sorry. So we can see the data like this and then check the data. So orderer have to reject if there are issues or upload it if everything is in order. This completes the overall workflow of R-CDE. Approve.

[SPEAKING JAPANESE]

Next, let's talk about the architecture.

[SPEAKING JAPANESE]

R-CDE adopt a microservice architecture consisting of three services.

[SPEAKING JAPANESE]

The platform service managed core data such as file, user, organization, site, and contract.

[SPEAKING JAPANESE]

It also provides a public API, which will be explained later.

[SPEAKING JAPANESE]

So file processing service managed time consuming process. One of these is heatmap generation. As mentioned earlier, the heatmap is generated by calculating the height differences between the scanned point cloud and design data, creating a point clouds to represent this difference in color. Since the point in the scanned point cloud are not evenly distributed, this process involves sampling necessary points before calculating the difference.

[SPEAKING JAPANESE]

The file processing service also handled point cloud LOD processing.

[SPEAKING JAPANESE]

Point clouds are too large to manipulate directly in our web browser. Therefore, as shown in the video, we display high-density point cloud closest to the camera and low-density point cloud far away. To achieve this, we divide the upload point cloud, integrate, and generate data with different density levels. We call it LOD data.

[SPEAKING JAPANESE]

Then there are the training processes.

[SPEAKING JAPANESE]

This is a stretch for the feature that are trimming of point cloud to arbitrarily oriented.

[SPEAKING JAPANESE]

Lastly, there's noise reduction processes.

[SPEAKING JAPANESE]

Point cloud inevitably include noise due to object or dust at the site during scanning. Therefore, we've implemented a process to remove data that seem irrelevant to the evaluation that is noise, by comparing design data with point cloud data.

[SPEAKING JAPANESE]

Finally, there is a smart contract service. As R-CDE include reliable in the name, all data is ultimately stored on the blockchain, making it difficult to tamper with. This service includes a private blockchain and handle read-write operation to it.

[SPEAKING JAPANESE]

The transaction on this blockchain are viewable by all users, like this.

[SPEAKING JAPANESE]

Now, let's discuss external integration.

[SPEAKING JAPANESE]

R-CDE provide mechanism for using external services and for external services to use R-CDE.

[SPEAKING JAPANESE]

R-CDE integrates API provided by external services, for example, to collaborate with photogrammetry and drones. For instance, using certain companies' API, where hundreds of images are uploaded to R-CDE at once, 3D data generated from these photos is stored in R-CDE.

[SPEAKING JAPANESE]

Conversely, R-CDE publish APIs for external service application and device to use. Data captured by 3D scanner can be sent directly to R-CDE from the device without needing to upload via browser.

[SPEAKING JAPANESE]

Let's also briefly discuss the infrastructure.

[SPEAKING JAPANESE]

R-CDE utilizes data risk. One distinctive feature is our use of Kubernetes for server orchestration.

[SPEAKING JAPANESE]

We use Kubernetes because R-CDE has many heavy process functions, as mentioned in the file processing service discussion. Therefore, we need to be able to spin up servers when needed and shut them down when not in use.

[SPEAKING JAPANESE]

When request for time consuming process like heatmap generation or trimming is received, the main server use Kubernetes to spin up servers for these heavy processes. These servers automatically stop after completing their tasks, ensuring no unnecessary costs are incurred.

[SPEAKING JAPANESE]

All right, this is Autodesk University. Let's talk about the most important topic. I talk about ACC integration from now.

[SPEAKING JAPANESE]

In R-CDE, we've made it possible to handle Revit data at design data, as you can see.

[SPEAKING JAPANESE]

When your user upload a Revit file to R-CDE, it's automatically uploaded to Autodesk Docs, where it's converted to SVF format.

[SPEAKING JAPANESE]

Currently, we're using Docs managed by R-CDE. But in the future, we aim to integrate with users own docks.

[SPEAKING JAPANESE]

R-CDE user callback API to be notified by Docs when the conversion is complete.

[SPEAKING JAPANESE]

Upon receiving this notification, R-CDE use API to retrieve the SVF and then process it to generate a SQLite database file and a GLB file, which are easier to use in web and Excel environment.

[SPEAKING JAPANESE]

In our Series Viewer, we use GLB file to display 3D models and SQLite to display properties.

[SPEAKING JAPANESE]

Yes, some of you in the audience might be wondering, why not use the SVF file directory? Why not use APS Viewer API?

[SPEAKING JAPANESE]

For my final part, let me explain why we are not using the APS Viewer API.

[SPEAKING JAPANESE]

Initially, we intended to use APS Viewer. But we encountered three challenges when trying to use the APS Viewer API in our service. Plus, it was difficult to compare different file. R-CDE is an application that need to generate data by comparing file, like point clouds and design data. While there are usually a lot of non-information or open source library available for common file types as well as Autodesk's proprietary format, making it difficult to create the processing mechanism for it.

Second, the Three.js user internally in the APS Viewer API differs slightly from the standard Three.js, making it impossible to apply library or techniques usable with the latest Three.js. R-CDE use React browser front end. Recently when handle Three.js in React, developers often use a library called React [INAUDIBLE] to simplify development. But this wasn't possible.

We also face challenges in incorporating ray tracing library and other Three.js-related issues. With their abundant documentation for popular 3D viewer library like Three.js or Babylon.js, there isn't much reference material for APS Viewer API. This made it time consuming for our engineer to catch up. And some expressed that they couldn't achieve what they want to do.

[SPEAKING JAPANESE]

Therefore, in our city, we established a mechanism to convert files to popular formats used by web engineers, creating a situation where front end engineer could freely develop the viewer using these files. However, this was just a mismatch R-CDE. And the APS Viewer is still convenient for creating views of Autodesk product. We hope to continue using it in various way and look forward to further evolution of the APS API. That's it.

[SPEAKING JAPANESE]

OK, thank you very much, Masamiki. And we also would like to discuss with Autodesk team this point. And, also, actually, we develop this system through the continuous development and through the every week meeting. And that is why I also would like to say thank you very much to Masamiki for your effort as well. OK, so the last part is open ecosystem for social implementation.

This part's keyword is open and industry wise. What I would like to say is not only for one company's profit. As I said, I work for Shimizu Corporation. But R-CDE is not only for the Shimizu Corporation. The profit provided by R-CDE should be shared among the construction industry.

This project itself is recognized as a activity in the area of cooperation, collaborated, and open. That is why we organize one working group for implementation. More than 40 companies and 80 participants from various industry joined this working group and conduct some activities with us.

This participant and the company is not only contractor but also client startup, software vendor, and maker. Regarding the contract, they are doing the pirating R-CDE on site. Of course, not only one site but also several sites. Regarding the vendor startup and maker, they conduct integrating their own service with R-CDE by their own effort. And the result of the integration, we utilize this integrated system in the site by a contractor.

When we consider the implementation, we also need to consider the scope of the R-CDE. Actually, the current scope of R-CDE is for Earthwork and NATM regarding the work type. And, also, regarding the production process, currently, R-CDE's scope is only for quality, as-built, and inspection.

So it means, of course, we need to expand this scope to other fields as well. That is why we need the strategy. So we have an open strategy for implementation. What I mean is we make current R-CDE as open source. And if we can make this R-CDE as an open source [INAUDIBLE] party's vendor or engineer can create the new application for other purpose, for example, for concrete work, cut and cover, quarter contour, and project progress measurement.

Of course, not only for the open source-- we need to develop the OPEN API service. So this OPEN API service include not only External API, which presented by Masamiki. But also we include the public API as well.

Actually, this strategy is still under the consideration. So today I cannot speak that [INAUDIBLE]. But if you are interested in-- please let me know. Anyway, we are still halfway through the implementation. So that is why we would like to continue this project to forward our project, to realize the implementation.

OK, that's all for our presentation. As I said, our journey is still undergoing. So if we get some progress, we are very happy-- we are very happy to share the progress as well in near future. Thank you very much for your attention.