No Lost Robots

By Andrea Er, Associate Software Engineer, GovTech

We can expect more robots in our lives in the near future. Therein lies a major challenge — keeping track of all of them. Wouldn’t want lost robots, would we?

I’m Andrea Er, an engineer with GovTech, and I’m part of the team developing a cost-effective solution for locating autonomous robots in multi-storey environments, such as hotels and office buildings. Our research is part of the Open Digital Platform, whose findings contribute to building Singapore’s Smart City infrastructure.

Our research has taken us to JTC Summit, a 31-storey office building in Jurong. We’ve been working with Jurong Town Corporation (JTC) to develop a software-based command and control system to manage a fleet of autonomous robots working in the building.

The system will give JTC better operational control and awareness of robotic operations within the building, and make it easier to introduce new robotic systems to the building’s various environments. It’ll accomplish all that without JTC having to install sensors or make other significant alterations to the building.

Matching multiple maps

Most autonomous robots are able to build a map of the environment they work in, a Robot Operating System (ROS) Map. The way a robot generates a ROS Map is no different than you wandering up and down your street, and sketching a map of the neighbourhood.

If you were then given a map of Singapore, locating your neighbourhood and your position in it is much more challenging. You’ll have to rotate your neighbourhood map to fit the orientation of the larger map.

Say you were also trying to locate all your friends in Singapore, and all they’ve given you are their respective neighbourhood maps. Now, you’re saddled with the task of matching all their maps to the big map of Singapore, rotating them, perhaps even changing their dimensions to make them fit.

That’s what we were figuring out how to do — how to align many individual Robot Operating (ROS) Maps with the floor plan of the building the robots are working in. That way, we’ll know the exact location of the robots within the building.

Using Real-World Coordinates

To synchronise the orientation of the two maps, the research team proposed using Real-World Coordinates to provide a common reference point. Real-World Coordinates are a set of map position values derived from global reference points.

Using Real-World Coordinates, each individual ROS Map is mapped onto the building’s floor plan. The heart of the research is coming up with a mathematical model which uses these Real-World Coordinates to rotate and transform the ROS Map onto the building’s floor plan.

With multiple ROS Maps and levels of the building successfully mapped, we then have enough data to generate a virtual 3D model of the building, and pinpoint the locations of all the robots within that model.

As you can imagine, there are many complexities in coordinating multiple ROS Maps in real-time, and the amount of map data to manage can be sizable depending on how large the building is and the number of robots working in it. Ideally, our system should be easily scalable for larger applications.

We were also concerned with how accurate the system is in locating the robots within the building. Errors in the individual ROS Maps, for example, will compromise the precision of the system. Ensuring the quality and accuracy of the ROS Map data is something we needed to verify at the earliest possible juncture, before they are fed into the system.

No Lost Robots

I’m happy and relieved that the system we’ve built has worked so far in our tests.

Tracking the movement of one autonomous robot in JTC Summit, we successfully followed the robot’s path through the building, even across multiple floors. Our mathematical model was able to align the robot’s ROS Map with the building’s floor plan, and we were able to visualise the robot’s movements within a virtual 3D model of the building.

The implications of our research findings are huge. If a software-based tracking system for multiple robots is feasible just by using Real-World Coordinates, then implementing robotic systems becomes so much more viable across a wide range of environments.

There would be significant cost-savings since no modifications to physical infrastructure or the robots are required. And we’ll be able to introduce, operate and track robotic systems from different vendors in multi-storey environments.

I’ve been a problem-solver most of my life, and engineering is my life’s calling. Solving the problem of lost robots has been so fulfilling, and I’m thrilled to contribute to a smarter future where autonomous robots will be a mainstay in our lives.

Wish to find out more about us or share your bright ideas on the possible use cases with us? Write to us at



SNSP develops whole-of-government platforms and solutions to sense, contextualise and act on 360° real-time sensor data. SNSP enables agencies to enhance planning, operations and service delivery to improve citizens’ quality of life.

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store