13 partners, 18 months, one mission: our consortium-led Transform-ER project aims to enable one million home upgrades every year by 2030. This blog series tells the stories of our tactical team of partners and the retrofit barriers they are addressing – so let us introduce Planarific, our AI accelerators.
So far, we’ve met Ambue, Energiesprong UK and Ultrapanel to learn more about their approaches to data, collaboration and offsite construction respectively.
For our next instalment, we’re chatting to Ran Xiao, founder and CEO of Planarific to hear how it’s leveraging machine learning to improve building stock assessments.
As AI and digital experts, they are developing tools to efficiently and cost-effectively assess portfolios, match typologies to solutions, and accelerate the process of surveying and costing to help our mission of achieving industrialised retrofit at scale.
Planarific is an AI-driven software company that accelerates assessment surveys and the design process for sustainable home improvement.
Our team focuses on scaling retrofits by transforming imagery from drones into semantically enriched 3D geometric models for 1000s of homes at a time. This means that metadata or labels are attached to the 3D model to differentiate each part of the building such as roof, wall, window, door.
For this, we are using our proprietary technology to deliver these models at an affordable price. And we can combine architectural insights with machine learning to provide analytics that help inform and support landlords, designers and contractors to plan and deliver retrofit programmes.
Our involvement in Transform-ER is driven by the desire to improve efficiencies and achieve economies of scale in retrofit. There are 1000s of the same typologies across the country and I was attracted to solving a problem for something that repeats a lot.
As a trained architect, I’m also interested in how human designers interact with an automated system. My PhD was about applying machine learning to architectural design and my belief is that these interactions are very important.
The UK needs to retrofit millions of homes, but the process remains slow, expensive, and fragmented. One of the biggest barriers is the lack of standardised, scalable solutions for assessing building stock efficiently.
Our work package focuses on overcoming this challenge by identifying architectural repetitions across portfolios, enabling batch processing rather than one-off assessments.
We are working closely with Ambue to integrate our rich visual and geometric data into a pipeline development platform to help landlords, manufacturers and contractors make informed, data-driven decisions. This will help create the market and demonstrate volume, while removing bottlenecks.
Ambue is looking at all the available data that’s available on a property and building this into the platform. Combining this with our unique level of insights into the geometry and visual features of a building, it has the potential to be revolutionary for landlords because there’s nothing else like it.
We’ve recently carried out a test drone survey of 500 homes in the Becontree Estate in the London Borough of Barking & Dagenham. This was to determine the logistics of collecting the data.
We would now like to expand this survey to a larger area of the estate to collect data from 2000 – 3000 homes. And we’re working with other landlords to carry out surveys on their stock so we can test the technology on a variety of typologies.
What’s interesting about developing software is that it’s an iterative process. You never really finish; you just keep iterating and making the product better for the customer.
Even though we're a tech focused company, we can clearly see that in this industry the human factors are incredibly important. For example, with our recent drone surveys it was important to communicate clearly to the community what we were doing and why. So, we are continuously learning about how our operations, as well as the retrofit process itself, impacts the people living in the homes.
It’s also important for us as a company that’s developing systems with AI that our outputs can be trusted and verified. To achieve this, we integrate a mechanism called ‘Human-in-the-loop’ which gives the human expert the agency to verify or challenge a decision. As I previously mentioned, the interface between human and computer is very important and must enable this kind of feedback flow, especially with something like construction. In practice, this means incorporating subtle design challenges when designing the software that presents the information in a way that allows the human to carry out checks.
For housing associations to invest in digital infrastructure now that will lead to savings in the long term. By this I mean spending some time and money on collecting better data and for us to help them use that data in a way that could help them lower their operational costs on either retrofit or related activities like maintenance and repairs.
For example, sometimes a retrofit programme is dictated by emergency repairs and the cost for this is taken from a retrofit budget. But our data could help anticipate the repairs that might be needed because we can carry out a roof survey as part of the data collection. So future proofing your data while future proofing your homes!
Find out more about Planarific at https://planarific.com/
You can sign up for updates on Transform-ER via the mailing list at the bottom of the project page. Stay tuned for our next blog in the Meet the Transform-ERs series.
