IoT Bridge Can Revolutionize Metal 3D Printing for Large Infrastructure

Autodesk and MX3D wish to improve the designs of future 3D-printed objects.

Engineers at MX3D and Autodesk are getting just a little bit closer to creating a science fiction landscape of organic-looking infrastructure.

With the invention of 3D printing robots that crawl along their creations, engineers are now able to build designs once thought impossible to make. One of the first is the MX3D Bridge being built in Amsterdam.

Stress lines from a simulation of the MX3D Bridge. (Image courtesy of MX3D.)

Stress lines from a simulation of the MX3D Bridge. (Image courtesy of MX3D.)

But there are a few catches. Though we can build these organic monuments, 3D printing robots aren’t exactly the fastest worker bees. Furthermore, since this is a new application of 3D printing technology, confidence in simulations that typically deem infrastructure safe need significantly more physical prototyping before they gain universal acceptance.

“For something as radical as this new production technique, we don’t have a lot of good simulation for it,” agreed Alec Shuldiner, Internet of Things (IoT) researcher at Autodesk and project manager of MX3D Bridge. “The design of the bridge is extremely complex, fluid and not a standard shape.Additionally, there are variations in the material and thickness output by the 3D printing robots. It’s hard to have a lot of confidence in the simulation. The fundamental problem is you can’t simulate things that are novel beyond what the simulation can represent.”

But this is when the MX3D Bridge team decided that the best way to learn what is really happening to their creation is to use the IoT.

Optimizing Metal 3D Printing From an IoT-Enabled Prototype

The project started with a prototype built at Autodesk’s Pier 9 CNC workshop. To this date, the prototype has gathered over 2 terabytes of data since February 2017.

“The prototype has been using a series of accelerometers, strain gauges and environmental sensors for CO2, light gain and sound levels,” said Shuldiner. “We’ve learned a lot about how to do modal analysis to place sensors and the number of sensors required, how clear your signal will be and how to handle such a large stream of data.”

Building a physical prototype of the MX3D Bridge. (Image courtesy of MX3D.)

Building a physical prototype of the MX3D Bridge. (Image courtesy of MX3D.)

To accommodate the data influx, Shuldiner explains that they have been using an Autodesk research technology dubbed Data 360. To then visualize this data, the team used another research technology called Dasher 360.

“We’ve recently hooked up Data 360 to Fusion Connect for when we get to the point to perform some actuation on the other end,” said Shuldiner. “Right now, we are collecting data and running machine learning models on it. The goal will be to have the system know how many people are on the bridge and where they are without using any cameras.”

With this information, the bridge can assist in the development of smart cities. Engineers will be able to use it to devise heat maps of population density within the city to optimize planning and improve people flow.

Additionally, designers can feed back this information to better describe their loads. This data can significantly help in the design of the bridge.

“One idea is to see if we can note the parts on the bridge that are extraneous or if the bridge was underbuilt,” noted Shuldiner. “Can we change the bridge in certain places and serve the same function with acceptable safety marks? That is the future of MX3D, using data that will inform their future metal additive manufacturing not just for bridges, but anything created using their metal 3D printing robots.”

The prototyping team started their research small and only put a handful of sensors on their bridge. Shuldiner notes it was a dozen strain gauges and an accelerometer located on each of the nine truss plates. The team found that the amount of strain differed based on how close a load, or person, was to the sensor.

This is to be expected. Now the question is, can they use this information, and data from other sensors, to teach a machine learning algorithm how many people were on the bridge and where? Well that was where their CO2, light sensing, cameras (with computer vision algorithms), temperature, air quality and a few other sensors came into play.

These sensors might not all make it to the final bridge in Amsterdam, but it could be used to help train the machine learning algorithm that will detect how many humans are on the bridge and their locations based on strain, weight and vibration.

“For example,” explained Shuldiner, “say we think we measure occupancy because we see a vibrational pattern on the bridge.We can also correlate this to increase CO2, which would suggest humans are breathing near the bridge. Or, it might be something the bridge is picking up nearby. You can imagine in Amsterdam there might be activity like a heavy truck that will register like a crowd of people crossing the bridge if the machine algorithm isn’t taught properly.The general concept of machine learning is you need corroborating data streams that are very different and not rely on than your primary signals.”

Shuldiner said that this exercise has two other benefits. First, it helps the MX3D team determine how to place the sensors and which sensors to use. For instance, Autodesk has learned a lot about how to use modal analysis simulations to determine optimum locations for the sensors. Second, they are also learning if the top-of-the-line fiber optic strain gauges will be needed or if something off the shelf will do the job nicely.

So Where Is the MX3D Bridge Now?

The answer is, not that far to be honest. 3D printing robots are not the fastest construction workers on the planet. As of July, the bridge in Amsterdam was a quarter of the way completed. As of now, it’s about a third of the way done.

(Video courtesy of The Times and Neil Bowdler.)

The team is also looking to develop ways to wire up the sensors needed to make the Amsterdam bridge the IoT success they intend to make it. Both Autodesk and MX3D decided to step away from wireless sensors due to the electromagnetic interferences in both the Pier 9 lab and the elements outside in the Red Light District. As a result, the wiring in the bridge will be essential.

In many ways, like most IoT products today, the idea to add sensors to the MX3D bridge was an afterthought. As a result, the team will need to modify the bridge after it is printed to handle the equipment.

Visual matrix of the MX3D bridge. (Image courtesy of MX3D.)

Visual matrix of the MX3D bridge. (Image courtesy of MX3D.)

However, this labor will bear fruit as Autodesk and MX3D will use their IoT data to better design metal 3D-printed products and IoT products in the future.

“Now with the IoT becoming more common,” noted Shuldiner. “Autodesk has tools to better anticipate and provide sensors.”

Written by

Shawn Wasserman

For over 10 years, Shawn Wasserman has informed, inspired and engaged the engineering community through online content. As a senior writer at WTWH media, he produces branded content to help engineers streamline their operations via new tools, technologies and software. While a senior editor at, Shawn wrote stories about CAE, simulation, PLM, CAD, IoT, AI and more. During his time as the blog manager at Ansys, Shawn produced content featuring stories, tips, tricks and interesting use cases for CAE technologies. Shawn holds a master’s degree in Bioengineering from the University of Guelph and an undergraduate degree in Chemical Engineering from the University of Waterloo.