
5.10.23-Ecosystem-Lamarr

-
Interactive transcript
NORHAN BAYOMI: Good afternoon, everyone. My name is Norhan. I'm a co-founder and COO at Lamarr.AI, and I also got my PhD from the Building Technology at MIT last year. And today I'm presenting Lamarr, which is a product of a research project originally that was founded in the Building Technology program between MIT, Georgia Tech, and Syracuse, and now we're commercializing the technology.
So climate change is posing a lot of risks to the built environment, with an increase in magnitude in climate events that would pose a lot of risk to the built environment, and the question that remains is how we can invest in the performance of the built environment in a very smart way to increase their performances and compact climate change impacts. So the Department of Energy projects that there will be 10 million buildings to be added to the US building stock within the next 10 years, so that would mean that we're expecting a new building stock to be performing with a high efficiency under the various risks that we're seeing from climate change. In addition to that, the policy is also heading with a direction to increase the performance of buildings, with new building codes, like Local Law 97 in New York and BERDO 2.0 in Boston, that would require mandatory reporting for emissions and energy use for buildings each year.
But also, the market is heading in the high-performance direction. So 61% of the construction projects that took place last year were retrofitting projects. Also, it's expected to double in size for building retrofit industry.
But the problem on the ground is much bigger. So 50% of the existing US building stock is built before the year 1980. So these buildings are performing with very, very low efficiency, and they would require to report and improve their energy use and emissions within the next years to combat climate change and energy, also, requirements. But also, there's a large potential to improve energy use in buildings with improving just the performance of building retrofit, as they can contribute to around 40% of energy losses that we can save by just improving building envelope, the facade and the roof performance.
So the problem with this right now is it's a very complete manual process that would require to hire a building auditor that would take all the images manually or even fly a drone, and then report the analysis and the information to the stakeholders, which, a process could take between one month to three months for an average building of 20,000 square foot. In addition to that, the picture you see in the middle is one of the tasks that's required to identify where are the leakage from the building, that sometimes you need to evacuate the entire building to be able to get accurate results.
Also, on the analytical side, all the analytical available right now are completely manual, so you have to go to an expert building scientist who would look at all the data and construct a digital twin with energy simulation to give you some sort of information about the return of investment by improving performance. So what we did in Lamarr is that we developed an AI software that uses information collected from drone or handheld thermal camera, and we, with our proprietary computer-vision machine learning, we can detect these anomalies fully autonomously.
How our technology works, it consists of three components. The first one is the data collection with drones that we have been developing over the years to identify the right path to collect both RGB and infrared data. And then all the data are processed into two different algorithms, one that looks at the anomalies in the thermal spectra and the second one to link these anomalies to building elements, which is, in that case, with a building corner or a roof or a slab. And then that classifies these anomalies into different types, and all these anomalies are registered into a digital 3D model that we construct from the 3D drone data, that could be used for energy simulation. And this entire process is fully autonomous, so we actually cut down the time that's spent in doing all this process manually.
So our service kit right now, we are providing five key components, from flight planning, data collection, data processing, geometry construction, and 3D energy modeling that's completely autonomous in our platform. This is an example of a couple of pilots that we did working with the MIT campus and Georgia Tech campus. This building's specifically from the Georgia Tech campus, where we tested the entire workflow on one of their buildings, so up to the construction of the 3D model, the registration, the anomalies, the shapes that you see here in yellow and red. And now we have a working better version from our software where you can actually upload your data, whether it's captured by a drone or a handheld thermal camera or a mobile camera, and that model can run and tell you exactly where are the locations of the anomalies, what type of locations, and give you some recommendations in a matter of seconds.
So it's a very fast algorithm. It can process 1,000 images in one second, and we have verified the performance and accuracy of the model with expert study of 50 building scientists from MIT and Georgia Tech. So through that pilot, we have identified that we can cut down the costs and time tremendously compared to the manual process.
And this is also another example from a pilot that-- a client we have right now in Florida, and you can see the outputs from the model. The thing that I want to highlight here is the last column in the table, which basically gives you a range of energy saving from HVAC and energy saving by improving these retrofits, improving these anomalies with retrofits. And we can also benchmark buildings against each other. So that was a group of buildings in the same setting, and we identified that there are a couple of buildings that are performing much, much lower than other ones. And we can cluster them and get some sort of a priority, which building that you need to intervene first.
So our team has the right skill set from building technology, AI, machine learning, and computer vision analytics. We span across MIT and Georgia Tech, but we're mainly based in Boston. And we have right now five full-time engineers in machine learning and building science.
And finally, I just want to reveal why we name the company Lamarr. It's a tribute to the actress Hedy Lamarr-- she's the inventor of the signal hopping technology-- but also a tribute to all inventors who didn't gain prominence during their time, especially women.
Finally, we're looking for collaboration, partnership with specifically large property management industry who are looking to improve and enhance the process of building inspection and performance. Also, we're looking to test our software with more experts in building science and property management. So if you are interested, please come to our booth and talk to us. Thank you.
[APPLAUSE]
-
Interactive transcript
NORHAN BAYOMI: Good afternoon, everyone. My name is Norhan. I'm a co-founder and COO at Lamarr.AI, and I also got my PhD from the Building Technology at MIT last year. And today I'm presenting Lamarr, which is a product of a research project originally that was founded in the Building Technology program between MIT, Georgia Tech, and Syracuse, and now we're commercializing the technology.
So climate change is posing a lot of risks to the built environment, with an increase in magnitude in climate events that would pose a lot of risk to the built environment, and the question that remains is how we can invest in the performance of the built environment in a very smart way to increase their performances and compact climate change impacts. So the Department of Energy projects that there will be 10 million buildings to be added to the US building stock within the next 10 years, so that would mean that we're expecting a new building stock to be performing with a high efficiency under the various risks that we're seeing from climate change. In addition to that, the policy is also heading with a direction to increase the performance of buildings, with new building codes, like Local Law 97 in New York and BERDO 2.0 in Boston, that would require mandatory reporting for emissions and energy use for buildings each year.
But also, the market is heading in the high-performance direction. So 61% of the construction projects that took place last year were retrofitting projects. Also, it's expected to double in size for building retrofit industry.
But the problem on the ground is much bigger. So 50% of the existing US building stock is built before the year 1980. So these buildings are performing with very, very low efficiency, and they would require to report and improve their energy use and emissions within the next years to combat climate change and energy, also, requirements. But also, there's a large potential to improve energy use in buildings with improving just the performance of building retrofit, as they can contribute to around 40% of energy losses that we can save by just improving building envelope, the facade and the roof performance.
So the problem with this right now is it's a very complete manual process that would require to hire a building auditor that would take all the images manually or even fly a drone, and then report the analysis and the information to the stakeholders, which, a process could take between one month to three months for an average building of 20,000 square foot. In addition to that, the picture you see in the middle is one of the tasks that's required to identify where are the leakage from the building, that sometimes you need to evacuate the entire building to be able to get accurate results.
Also, on the analytical side, all the analytical available right now are completely manual, so you have to go to an expert building scientist who would look at all the data and construct a digital twin with energy simulation to give you some sort of information about the return of investment by improving performance. So what we did in Lamarr is that we developed an AI software that uses information collected from drone or handheld thermal camera, and we, with our proprietary computer-vision machine learning, we can detect these anomalies fully autonomously.
How our technology works, it consists of three components. The first one is the data collection with drones that we have been developing over the years to identify the right path to collect both RGB and infrared data. And then all the data are processed into two different algorithms, one that looks at the anomalies in the thermal spectra and the second one to link these anomalies to building elements, which is, in that case, with a building corner or a roof or a slab. And then that classifies these anomalies into different types, and all these anomalies are registered into a digital 3D model that we construct from the 3D drone data, that could be used for energy simulation. And this entire process is fully autonomous, so we actually cut down the time that's spent in doing all this process manually.
So our service kit right now, we are providing five key components, from flight planning, data collection, data processing, geometry construction, and 3D energy modeling that's completely autonomous in our platform. This is an example of a couple of pilots that we did working with the MIT campus and Georgia Tech campus. This building's specifically from the Georgia Tech campus, where we tested the entire workflow on one of their buildings, so up to the construction of the 3D model, the registration, the anomalies, the shapes that you see here in yellow and red. And now we have a working better version from our software where you can actually upload your data, whether it's captured by a drone or a handheld thermal camera or a mobile camera, and that model can run and tell you exactly where are the locations of the anomalies, what type of locations, and give you some recommendations in a matter of seconds.
So it's a very fast algorithm. It can process 1,000 images in one second, and we have verified the performance and accuracy of the model with expert study of 50 building scientists from MIT and Georgia Tech. So through that pilot, we have identified that we can cut down the costs and time tremendously compared to the manual process.
And this is also another example from a pilot that-- a client we have right now in Florida, and you can see the outputs from the model. The thing that I want to highlight here is the last column in the table, which basically gives you a range of energy saving from HVAC and energy saving by improving these retrofits, improving these anomalies with retrofits. And we can also benchmark buildings against each other. So that was a group of buildings in the same setting, and we identified that there are a couple of buildings that are performing much, much lower than other ones. And we can cluster them and get some sort of a priority, which building that you need to intervene first.
So our team has the right skill set from building technology, AI, machine learning, and computer vision analytics. We span across MIT and Georgia Tech, but we're mainly based in Boston. And we have right now five full-time engineers in machine learning and building science.
And finally, I just want to reveal why we name the company Lamarr. It's a tribute to the actress Hedy Lamarr-- she's the inventor of the signal hopping technology-- but also a tribute to all inventors who didn't gain prominence during their time, especially women.
Finally, we're looking for collaboration, partnership with specifically large property management industry who are looking to improve and enhance the process of building inspection and performance. Also, we're looking to test our software with more experts in building science and property management. So if you are interested, please come to our booth and talk to us. Thank you.
[APPLAUSE]