Interview

Looq to the future

Lukas Fraser, Vice President, Looq AI, in conversation with Darrell Smart 

Lukas Fraser talks to Geospatial Engineering about bridging the gap between the physical world and the digital

ESTABLISHED in 2021, Looq AI is a high-growth technology platform company dedicated to advancing critical infrastructure digitisation and diagnosis. It has developed new technology – using camera AI technology – to simplify the process of generating geometrically precise, geo-referenced and intelligent digital twins to empower utilities, engineering, surveying and construction ecosystems to build and uphold a safer, sustainable and electrified future.

Bridging the gap between the physical world and the digital, the company is enabling true asset intelligence of critical infrastructure, rendering undocumented infrastructure a thing of the past.

The technology is delivering the information that you need, when you need it and at a scale that matters. Civil Engineering Surveyor’s Darrell Smart caught up with Looq AI’s vice president, Lukas Fraser to find out more about the company and how it is changing the way in which surveyors and engineers are quickly and safely collecting and processing, large-scale accurate data.

Can you tell me a little bit about yourself, how you’ve got to where you are today and about Looq AI itself?

I studied geomatics engineering at the University of New Brunswick with the aim of becoming a land surveyor. That was my goal – I liked the thought of working in the woods and being outside. After graduating, I took a job as a field surveyor for an aerial mapping company. I got really interested in aerial mapping and I’ve been working in remote sensing ever since. I had the opportunity to work on some large lidar and imagery projects for the US government’s 3D Elevation Program. It was really cool to see these datasets become available for public use.

I strongly believe that good geospatial information is crucial to the success of key public infrastructure projects. After a couple of years I moved to San Diego where I was working at an engineering company, NV5, helping to kickstart its drone lidar programme. My career then moved more granular after transitioning from working with planes and helicopters to UAVs.

That was a big challenge taking the datasets, point clouds and imagery and turning them into useful information.

I met Dominique Meyer, CEO and co-founder of Looq AI, and we used to argue about the merits of lidar versus photogrammetry. In a fast-paced project environment, you don’t really get the chance to focus on any one thing for long – you’re often delivering one project and then you’re on to the next.

When the opportunity came to create something that could potentially not just change the outcome of one project, but the outcomes of thousands, I jumped at the chance.When the opportunity with Looq AI came up to focus and create something that could potentially not just change the outcome of one project, but the outcomes of thousands, I jumped at the chance. At Looq AI we don’t just want to work with point clouds or cool datasets, but create the tools that enable people to do their work in a fast, accurate and affordable way.

How did you come up with the concept of using camera AI technology?

Dominique and Shreyas Niradi, our chief science officer and co-founder of Looq AI, – along with a couple of other students – were studying for their PhDs at the University of California, San Diego, working in the lab together.

Their focus was self-driving car technology, working on AI and computer vision around automated mapping. That transitioned over time as they realised that the geospatial industry seemed underserved. That’s where the concept came from. It all started in a garage, from which we still keep the original robot. It’s on tracks and would travel through tunnels or buildings with a bunch of cameras on it mapping the area.

The Looq qCam in action.

What are your goals?

Our main goal is to create a simple, cost efficient and accurate way to capture, analyse and access 3D information. The land surveying industry is notoriously underappreciated. The tools that we have and the work that needs to be done is incredibly challenging. You need to have a lot of knowledge about coordinate systems within the world and you must know how to use many different pieces of technology. It can be extremely complicated to get from point A to point B in terms of a project.

The land surveying industry is notoriously underappreciated.We are a computer vision company at heart, but we decided that we needed to start from the ground up, which is how we ended up with our solution. The technology comprises of a camera, the qCam, with built-in GNSS; a cloud processing solution, qAI, and an app, the qApp, which is used for viewing, analysing and collaborating on exporting 2D and 3D information.

The hardware processing software application takes you all the way through the different phases of a project.

Tell us about the methodology

On a typical project, you walk around the site with the camera and once complete, the data capture from the camera syncs via the internet and processes the information on the cloud. I know that a lot of land surveyors care deeply about coordinate systems and projections and we have a lot of customisation. It’s complete within 24 hours and ready to be worked on. It’s a quite simple and easy way to carry out a survey with valuable information.

We are a computer vision company at heart, but we decided that we needed to start from the ground up, which is how we ended up with our solution.How does this differ from other methods of data capture?

It’s very different. Our hardware was designed specifically to go with our cloud processing and vice versa. Everything about the camera and our processing pipeline are tuned specifically to work with each other.

This can be seen in the camera overlap with its field of view designed to enable surveyors, engineers, for example, to go out and create accurate datasets quickly. There’s no tuning with parameters, it’s turned on and then with access to the stream, it’s ready to go.

It is corrected GNSS; so, as with other offerings it’s RTK or PPK quality depending on its location. We do also offer a PPP correction in remote locations without base station access. You can get 3-5cm global accuracy assuming it has an unobstructed view of the sky. The ranging or distance measurement precision accuracy is sub-centimetre and those are the numbers that we have been getting when testing against traditional methods like total stations.

There are other camera solutions out there, but for us it is the fact that we designed everything very specifically and intentionally with this end goal in mind – it’s a holistic end-to-end approach.

How does this type of technology benefit the industry?

Land surveyors are using it for topographic mapping, putting the collected data into software, such as Autodesk Civil 3D; creating surface files and making top-down images for drafting as well. We try and create workflows that people are familiar and comfortable with. So everything that land surveyors are used to seeing in land surveying software, we try to add tools to support that to increase its ease of use and adoption.

This can be seen in the camera overlap with its field of view designed to enable surveyors, engineers, for example, to go out and create accurate datasets quickly.There’s a big demand for pole loading analysis. Engineering companies – and surveyors – walk with the cameras, around 30-45 seconds per pole, and gather 3D information on these poles.

There is a heavy focus on its automation; extracting points of attachment on the pole and the lines on top of the pole in all the different modes and then feeding that into pole loading analysis software – information that has traditionally taken a long time or is difficult to obtain, is now easily achievable.

Safety is also a big driver for us. We are de-risking a lot of general project tasks by keeping operators out of high traffic areas.

One particular client in America needed to be able to quickly capture power poles and power lines in people’s backyards and, for them, it had to be able to be done anywhere in the country.

Shipping workers from California, for example, all over the USA would take too long. It meant shipping the cameras to its office locations to give its workers, who were able to watch training videos, an easier option to start collecting data without any logistical challenges. It was all processed on our cloud in one centralised location and the company is now able to go back out and collect more information as needed.

I’d like to believe that as someone who has done a lot of land surveying and mapping, that I know exactly what everyone wants. But that is not true. A lot of my time is spent talking to surveyors and engineers, understanding what they are doing and how it could be done better.

We have also worked on a couple of projects with substations on vegetation management. In California, the wildfires are a massive problem and vegetation management programmes are needed. Our technology is able to identify the power lines, power poles and vegetation – that information can then be fed into systems to help identify encroachments or fire areas.

In addition, natural disasters such as hurricanes, fires and mudslides can be responded to quicker enabling an accurate snapshot to aid disaster relief. What’s next for Looq AI? We are focused on the AI portion of Looq AI.

We have built an awesome piece of hardware; we have an automated way to create point clouds and imagery and now we need to turn these datasets into information.

Point clouds and images are great, but what’s next? We have a tonne of items on our roadmap and we have a lot of things that we want to accomplish.

I’d like to believe that as someone who has done a lot of land surveying and mapping, that I know exactly what everyone wants. But that is not true. A lot of my time is spent talking to surveyors and engineers, understanding what they are doing and how it could be done better.

It influences our roadmap decisions quite heavily. We welcome engineers and surveyors to help shape the decisions we make now and in the future. 

Lukas Fraser, Vice President, Looq AI, in conversation with Darrell Smart

www.looq.ai

@looq_ai