Sensing the Future: Exploring Novelda's Ultra-Wideband Radar Sensors

Welcome to another exciting episode of the CTRL+Listen Podcast! In this episode, hosts James and Nora dive into the world of cutting-edge sensor technology with special guest Fredrik Kervel from Novelda, a Norwegian company revolutionizing human presence detection using ultra-wideband radar sensors. Learn how Novelda's pulse radar technology works and its applications in various industries, from building automation to automotive. Frederik explains the intricacies of sensor decision-making, the importance of machine learning, and the future of sensor technology.

Listen to the Episode:

Watch the Episode:

Episode Highlights:

• Radar Use-Cases AI and
• Radar Sensors
• The Future of Sensor Technology
• Complimentary Sensor Systems
• How Sensors Interpret Data

Links and Resources:

For more information about Novelda's groundbreaking work, check out their website

For more about Fredrik, get in touch here

Transcript:

James: Hey everyone this is James from the control listen podcast by Octopart today we have a guest all the way from Norway for you it is Frederick Kervel from Novelda really interesting company does some amazing sensor technology thank you so much for coming on the show great to be speaking with you.

Fredrik: Thank you for having me it's exciting to be on your podcast and be able to talk about what we do in Novelda.

James: You do some exciting stuff do you want to tell us a little bit about the company and what it is I guess what your mission is what your product is?

Fredrik: Absolutely so as you said it's a a Norwegian company we're a spinoff from the University of Oslo from a research group there and our our CTO is still a professor at University and and very active there and we started making ultra wideband radar sensors specifically for human presence detection more than 10 years ago AG go and that's what we've been focusing on working on and and perfecting since since then so making sensors that are low power and able to detect human presence in a relatively confined area.

Nora: Wow so how does how does a sensor work at a basic level and for for detecting human presence what is it what is it used to detect human presence?

Fredrik: So it's we are a pulsed radar so it means that we we send a a single pulse from an antenna and it's very very short in time and wide in frequency hence Ultra wide band and that single pulse will propagate through space hit objects and then reflections will come back to the sensor so you can think of it like  how a bat navigates in darkness for example or these sonar pings that you may have heard in u-boat movies so we sent thousands and thousands of these pings these single pulses and with that we can build an image of the surroundings of the sensor and so we make frames as we call them which are single instances of time and over multiple frames we can very very accurately detect a a small movement even with some millimeter millimeter displacement and that's perfect for detecting a human person just sitting still breathing for example we can basically just see the the movement of the chest when you're sitting completely still and just breathing.

Nora: Wow but I noticed that you're saying human are animals different or are they also able to be detected?

Frederik: It's a good question and it's very relevant for lots of the applications where you want to detect humans but you may not want to detect animals and it's actually a part of what makes radar sensors complicated to use because you're right of course animals and humans are very similar so it's it can be quite difficult to to differentiate between them so the easiest way to do it is look at the the amplitude of the signals we get back so normally a person will not just appear out of nowhere and sitting completely still and breathing there will be some larger movement first right and those movement will be larger than what a cat or even a small dog will have interesting so we can differentiate on that and then we can also do stuff with just field of view where you're looking and so forth and there's more complex way to ways to do it but that requires more computing power basically.

James: I guess people probably would want to know what's the applicability of this like what would this be used for and how is it beneficial?

Fredrik: Yeah so that's that's  anything really and it's becoming more and more common in even applications that you would call relatively simple so for example light control in buildings turning on and off lights in rooms based on presence the common way to do that through sensors today is with a passive IR sensor and I think many people have noticed that if you're alone in a room or just a few people there and you're sitting quite still the lights will eventually go off and you have to sort of wave at the sensor to turn them on again and that annoyance you don't have with the radar sensor because we are much much much more sensitive and able to detect people even if they're still that's a simple application more advanced applications would be respiration monitoring so we're in a few products Sleep Quality monitors which you can put on your bedstand for example which will be able to detect your respiration pattern throughout the night baby monitors which detects the baby's respiration and also various types of Health applications where you want to observe elderly people for example remotely and yeah get the vital signs monitoring.

James: Wow and I know you mentioned when we spoke via email that you had I guess four key sort of products do you want to sort of talk through those and the differences between them?

Fredrik: Yeah so right now we're about to launch two products based on our current X4 sensors that's our X4 IC which we have designed and made and we put that on a complete sensor module and developed software which you get for free so that we have a turnkey solution so it's the proximity sensor and the occupancy sensor and the difference between them is that the proximity sensor is fine-tuned and designed for products where you want some kind of a human machine interaction think about the displays you want to enable when people approach them for example it could be ATMs could be  marketing displays whatever and the occupancy sensor is is designed and fine-tuned for  room occupancy light control mainly so you can integrate it into your luminaire into your LED panel and it will turn on the light basically when people are in the room.

James: so so you had it was the proximity sensor the occupancy sensor the ultra low power presence sensor and the UWB X7 dev kit.

Fredrik: Yes we are also developing and sampling our Next Generation products which is based on our new X7 IC so the main differences between the X7 and the X4 is that we have reduced the power consumption significantly we have added more computing power on the device and most importantly we've added another Radar Channel so we have two radar channels now which basically enables the sensor to do positioning of people inside the detection zone so it opens up for lots of more complex applications like imagine a meeting room where you want to not only see if there's people there but determine how many people there are and where they're sitting.

Fredrik: And with a multi channel radar you can do that so what we have in the works to begin with is what we call the ultra low power presence and we've already showcased that running continuous presence detection at about 50 microamps which is a lower current consumption than anybody else okay basically and we also have what we call radar direct which is the ability to control the radar directly through an API and get the raw radar data out so if you want to do your own signal processing and basically make your own algorithms in your product we enable that on the X7 device and as I said these products are currently in development and sampling so they're showcased on our web page and you can click on a request evaluation kit button.

Nora: This is interesting what industries utilize radar technology the most because obviously we're talking about like applications or you know security and just basic monitoring but I wonder are there radar applications like and factories or you know in space or just other other uses of radar technology that people may not be aware of.

Fredrik: Yeah so the applications we are targeting are mostly in consumer electronics for for commercial buildings building Automation and also inside the automotive industry we are not we don't have any any products in space unfortunately for sort of factory use there are use cases we've seen but we're not in there for example if you want to make sure that there's no people inside a sewn with restricted access it could be hazardous for example art type sensor can be used to to ensure that that area is empty we've seen it in other types of applications where you have not in factories but where you have hazardous areas and you have service Personnel for example that enter these zones from time to time and you want to make sure that when the system is operating in Normal Al mode it's it's there's no people in those areas and for for Automotive there's of course there's plenty of different radar applications you have everything from the radar is detecting the cars in front of you you have a kick sensor where you can move your feet underneath the car to open the trunk and and press detection for inside the cabin is also now coming so you have this hot car act in the US basically mandating a sensor solution to be able to detect kids or pets or yeah people left behind in the cars and this will also be enforced in in Europe so that's a quite large Market right there and a radar sensor is basically the the perfect solution to detect human presence inside a car for those applications.

James: I know that right now a lot of the vehicles to get around that use weight sensors and pick up so they can tell if it's like groceries on the seat versus a dog or a child type thing.

Frederik: Yeah and and also secondary to that or for carbon I guess it's is as important is  as you can do with weight be able to position people in the car for various reasons so the car systems basically want to know how many people are in the car and where they're sitting.

James: Interesting I was wondering for like a data analytics perspective I imagine this is extremely useful to monitor traffic in spaces and work out peak times like days of the week they actually have high activity that sort of thing.

Fredrik: Yes and that's one of the things that we are seeing that's becoming more and more popular and asked for especially in in building optimization or building automation to be able to determine utilization of office space for example see how efficient your office layout is if you have too much space which areas are used more which are used less and of course in addition to that be able to to control anything that has to do with power consumption so that you can not only control the lights but also HVAC systems and so on so you can dynamically control everything that goes on in the building based on how many people are there and where they are and for that you need more accurate sensors than what you get with a passive IR you need to be able to have some count of how many people  are in the building and also know where they are in real time.

James: I imagine it's the same for something like if you're doing foot traffic monitoring say in like a particular space like if it's entrance to a subway station like which entrance gets used the most versus other ones you can just mon that that way as well.

Frederik: Absolutely we are not in in any applications like that but definitely anywhere where you have people passing through some kind of a passage confined area it's it's possible to to collect data and do some analysis on it

James: That’s exciting so much so many possibilities of what this can be used for so applicable.

Frederik: Absolutely and it's also a challenge right because some of these applications are very challenging or complex to make a robust solution for so you have to pick the right one and or the right ones and make sure you make a a good solution for for those.

James: Do you see the advancement of AI as unlocking increasing amount of potential in this space?

Fredrik: Absolutely so with any sensor like like a radar sensor like ours where you do complex signal processing and also are able to collect lots of data machine learning is is a great Tool both to develop and optimize algorithms based on collect the data and also to run in real time on the single processing.

James: I guess I wanted to ask what do you see as sort of the future of sensor technology what where do you see this going is there unlock potential that there we just need to get there technology wise.

Fredrik: Yeah I mean you can basically see that everything is becoming more sensorized there's sensors everywhere already and the demand will just keep growing and we're also seeing that sensors get better performance they get more processing power on the sensors themselves so enabling them to do more complex stuff on the sort of edge nodes so with that you'll be able to basically sort a lot of more complex problems and making a more how should I say dynamic world like taking what we talked about for building automation and just scaling it up so that you can adapt the world around us more and better.

Nora: Interesting are there any instances of sensor systems that work that are that are taking different types of data and then working together because I thought of this because you know we have five senses that all work together to decipher information that is around us but you know a mechanical sensor is only getting one thing and sees yeah one interpretation is there a way to have like sensor systems?

Frederik: Absolutely and many sensors can complement each other or or work together both at the same time but also in sequence so we had the sensor integrated into laptops and the purpose was to be able to wake up the computer when some someone approach approached it and then locked the computer when the user went away and many pieces come with face recognition today and you could think that that same system could be used to detect the person but the problem was that it was consuming too much power so you needed something another sensor which was more low power to enable the second sensor that did the facial recognition and there's also other examples where you have for example a radar can propagate through most building materials so if you have in a normal home for example you have wooden walls or dry walls and the radar signals would propagate through these walls and potentially detect people in the neighboring room so to avoid that you could complement the radar sensor with something that would confine it to to the room so you'd get the responsiveness on the radar sensor but have another sensor that would make sure that you don't detect stuff that is outside the room for example and I mean there's with with non- radar sensors there's tons of other examples where you sort of complement sensor data.

Nora: Do sensors ever is there ever an issue with like how does a sensor make a decision or make a judgment?

Fredrik: Yeah I mean it yes I mean one of the things that we  want to avoid the most is false positives and false negatives so you don't want the sensor to turn on the light when there's no one in the the room and you definitely don't want the sensor to turn off the light when there are people in the room so that's a a sort of a simple example for making that decision and in a sort of simple occupancy sensors case it's just down to how you have built your signal processing and it depends then on on the tuning what kind of what kind of movements are you relying on to determine if there's people there or not if you make it too little sensitive it will turn off when there's presence if you make it too sensitive it could potentially trigger on something that's not a person for example so getting that balance right is something that we spend a lot of time on and making sure that we have systems and and collect sufficient amount of data for our products to ensure that they are as robust as they can be for the type of application we're we're working with.

James: I think there's a lot of I guess misconception from people on on what kind of constitutes a sensor and how common they are I mean they're in almost every part of people's daily lives at this point why do you sort of think some of the examples would help people realize how common senses are what are everyday objects they use in I wouldn't even think of?

Fredrik: Yeah I mean as you say it's it's it's basically everywhere and I mean you can start by your Smartwatch right it's very common now for people to have a smartwatch and on the Smartwatch I mean how many sensors do you have you probably have a temperature sensor right you have something that measures your heart rate you'll see air pressure basically anything  that goes on in the environment around you and also in in the body and that's just starting with what you wear then you go out in your car right and you have as we've talked about sensors that detect which seat you're sitting in and sensors that detect the environment around the car and sensors that detect if you want to open the trunk and come to the office and maybe there's a sensor there that opens the door for you and maybe you look at a display that realizes that you're there and it acts on that somehow and it just goes on and on and on and I mean maybe you have a weather station at home that's a bunch of sensors right there you have sensors that control your heating your air condition yeah so sensors are everywhere and and we'll probably just get more and more sensors in our devices.

James: I think it's great for people to understand that because I think I think people sort of when you say like I guess monitoring sensors or radar technology some people like well I don't really want to be monitored I don't want sensors around me well it's too late for that yeah it's it's everywhere.

Fredrik: It is absolutely so one thing that you can say about radar sensor is that there are a bit non-intrusive in the sense that you cannot identify anybody with a radar sensor right today you can only detect if there's someone there with some signal processing maybe you can classify it as a human as opposed to an animal or something else but there's no way you can take our technology and the data we generate to accurately identify anybody so it's non-intrusive in that sense as opposed to a camera for example where a an image is definitely something you can use for identification.

Nora: What kind of sensor is facial recognition?

Fredrik: So facial recognition is a is a camera and I believe they typically use these IR cameras which are same as when you have like night vision which the camera is just figured out.

Nora: So it's not using any kind of sensor technology?

Fredrik: No it's I think it's just image processing to measure like depth points on your face so it's a 3D visualization of the face with the camera.

Nora: One question that we like to ask all of our guests is if you had known what you know now before Covid how would that have changed how you approach technology and what you do?

Fredrik: Oh that's a good question I think for us for our products and our product development it's wouldn't have changed a lot actually I think as for many other companies it's more about how we work right and how it opened up for remote work one thing we learned a lot from is we were right in the start of integrating with large customer in Japan when everything shut down and that was of course extremely challenging not being able to to go there and be on site and help out so we basically had to scramble and and hire people in in several countries in in Asia and then also not being able to go there and meet these people face to face and that was a really interesting experience and we were actually able to to get some really really good people on board back then which we still have in the company that are now part of our sales and support organization in in Asia so yeah I don't think we would have sort of done anything differently but we've definitely learned a lot about how we work I think.

James: Great well we're kind of at time here so obviously we wanted to say thank you so much for coming on the show but I also wanted to just ask you if people want to support the company or look at your products or follow you on social media what are the best places for them to do that?

Fredrik: So the best place is to go to Novelda.com that's where you'll find all the information about our products and how you can get access to evaluation and development kits and how you can get in touch with us if you want to that is definitely the best entry point fantastic.

James: Great thank you so much for coming on the show.

Nora: Thank so much for talking to us.

Fredrik: Thank you for having me, it's been cool.

James: Anytime, and for anyone listening come back next week we'll have another guest for you.