It*#8217;s not often that you come across a geo-startup that has its roots in the PhD thesis of its founder, let alone be mentioned as one of the *#8220;50 most promising startups you*#8217;ve never heard of (2017)*#8221; by Bloomberg.* But that is not what makes*Navisens, an interesting company. What makes navisens a startup to remember is their technological prowess *#8211; their positioning API works with just the inertial sensors in your smartphone and all you need to do to actually see if it really works is check out their demo on their website. Yup, you don*#8217;t need an app for that! We spoke to Ashod, their CEO and founder to learn more about Navisens and their story. Read on!*Navisens Team at the Augmented World Expo (AWE)Q: Ash, let’s get straight into it. Your PhD, which is where the navisens story actually begins, was aimed at locating firefighters inside a multi-story building. What motivated you to do a PhD in this specific topic? *What were you doing before that?A:*The story of my PhD which led to the founding of Navisens begins with my undergraduate degree in electrical engineering where I became interested in sensors, real-time embedded software, and Machine Learning (ML). This led to my undergraduate thesis in robotics where I worked on an autonomous Unmanned Air Vehicle (UAV). I gained my first experience with inertial sensors (accelerometers and gyroscopes) and I designed the entire hardware and software system for communication, data acquisition, and control.My hard work and successful performance in my undergraduate thesis led to the offer of a scholarship for a PhD in robotics where I was based at one of the world’s leading research centres for field robotics, the Centre for Autonomous Systems. I not only completed my PhD, I also qualified to bypass the requirement to complete a Masters degree, so I went straight from undergrad to a PhD. During my journey, I worked as a researcher on projects which very few people in the world have the opportunity to work on, including the 2007 Defense Advanced Research Project Agency (DARPA) Grand Challenge where I was based at UC Berkeley.
In my PhD research, given my background in sensing, machine learning, and hardware and software design, I became interested in the localization problem which is a fundamental problem in robotics. Most indoor localization solutions rely on lasers and cameras. It’s not feasible to mount these type of sensors on humans as they*#8217;re heavy and cumbersome, consume a lot of power, and require an unobstructed field of view. This led me to focus on tracking firefighters indoors, which suited the challenge I had set for myself: to track a firefighter without placing any sensors in the environment and without collecting any data beforehand. This is what eventually turned into Navisens. Q:*Unlike many other (indoor) positioning solutions, Navisens does not rely on any additional infrastructure. All you need are the inertial sensors inside a smartphone. But then, how accurate are these sensors and How do you deal with drift and other engineering challenges that come with them?A:*Navisens doesn*#8217;t rely on any additional infrastructure and this originates from the background of Navisens tracking firefighters, where it’s not possible to install any infrastructure in a building. Due to that constraint, my PhD research was focused on using only self-contained sensors such as an Inertial Measurement Unit (IMU). Most robotics applications employ an IMU, and I had used several IMUs in many robotics research projects. The key point is that traditional inertial navigation algorithms rely on very accurate sensors. The IMU my team used for the 2007 Defense Advanced Research Project Agency (DARPA) Grand Challenge was ITAR restricted, meaning that the sensor quality is high enough that the U.S. government controls the export of the device as a military-related technology.Smartphones include mass-manufactured low-cost micro-electro-mechanical systems (MEMS) sensors, which are several orders of magnitude lower quality, and thus traditional approaches and algorithms cannot be used. Navisens handles this with our proprietary algorithms which process the sensor data in a unique way, very different to traditional approaches, and based on what I would call modern techniques such as machine learning and newer robotics algorithms. Our goal is to minimize the drift to be as low as possible such that our location accuracy is suitable for the majority of applications without having to use any external measurements or corrections. Q:*When you talk about a startup that works on positioning with inertial sensors, the obvious question that most people have (at least of late) is *#8211; Are you guys also working on autonomous navigation? You were listed in Bloomberg’s “These Are the 50 Most Promising Startups You’ve Never Heard Of”(congratulations), so needless to say that would have also meant that the automotive industry was at your door. So?A:*Thanks, it was great to be listed in the Bloomberg article… I actually didn’t know about it until it was published!We’ve certainly been approached by automotive manufacturers and OEMs, especially since my background includes a lot of robotics and autonomous vehicle research. In fact, as an academic researcher, I’ve also worked on projects designing sensors and software tools for vehicle dynamics analysis for automotive manufacturers. This puts me in a unique position to understand the field very well.Firstly, as a founder * CEO of a technology startup, one of the things I’ve learnt is the need to remain focused. Even small distractions or changes in direction can consume resources unnecessarily and can have an impact on your team. Staying focused means that your team isn’t spread thin and you can continue to service and delight your current customers and maintain quality work. In addition to all that, some of the very largest companies in the world (which includes automotive companies) can be a large resource drain on a startup. You always need to be mindful of this as a leader of a startup.Secondly, as an engineer, there are some significant differences between a vehicular platform and a smartphone-based platform. As Navisens provides a Software Development Kit (SDK) for smartphones, we avoid custom operating systems and an operating system which isn’t iOS or Android. This rules out certain vehicle integrations. Since we’re tracking smartphones which are carried by humans, we don’t have the benefit of a comparatively much more stable platform such as a vehicle, and we don’t have the benefit of knowing the control inputs (e.g. steering, brake, throttle) of a human. Thus, we’re solving an extremely challenging problem, one which requires us to have an extreme focus on technology.Considering the above, I will mention we are actively in discussions with one or more automotive companies in some capacity
*Q:*One of the coolest things that fascinated me the most about Navisens was your demo which is available both as an app (on Android and iOS) and on your website (on iOS devices). I didn’t even know that you could access the inertial sensor data from a website! The motivation for the app is clear but what made you build it into your website. Was it a bet with someone?
In my PhD research, given my background in sensing, machine learning, and hardware and software design, I became interested in the localization problem which is a fundamental problem in robotics. Most indoor localization solutions rely on lasers and cameras. It’s not feasible to mount these type of sensors on humans as they*#8217;re heavy and cumbersome, consume a lot of power, and require an unobstructed field of view. This led me to focus on tracking firefighters indoors, which suited the challenge I had set for myself: to track a firefighter without placing any sensors in the environment and without collecting any data beforehand. This is what eventually turned into Navisens. Q:*Unlike many other (indoor) positioning solutions, Navisens does not rely on any additional infrastructure. All you need are the inertial sensors inside a smartphone. But then, how accurate are these sensors and How do you deal with drift and other engineering challenges that come with them?A:*Navisens doesn*#8217;t rely on any additional infrastructure and this originates from the background of Navisens tracking firefighters, where it’s not possible to install any infrastructure in a building. Due to that constraint, my PhD research was focused on using only self-contained sensors such as an Inertial Measurement Unit (IMU). Most robotics applications employ an IMU, and I had used several IMUs in many robotics research projects. The key point is that traditional inertial navigation algorithms rely on very accurate sensors. The IMU my team used for the 2007 Defense Advanced Research Project Agency (DARPA) Grand Challenge was ITAR restricted, meaning that the sensor quality is high enough that the U.S. government controls the export of the device as a military-related technology.Smartphones include mass-manufactured low-cost micro-electro-mechanical systems (MEMS) sensors, which are several orders of magnitude lower quality, and thus traditional approaches and algorithms cannot be used. Navisens handles this with our proprietary algorithms which process the sensor data in a unique way, very different to traditional approaches, and based on what I would call modern techniques such as machine learning and newer robotics algorithms. Our goal is to minimize the drift to be as low as possible such that our location accuracy is suitable for the majority of applications without having to use any external measurements or corrections. Q:*When you talk about a startup that works on positioning with inertial sensors, the obvious question that most people have (at least of late) is *#8211; Are you guys also working on autonomous navigation? You were listed in Bloomberg’s “These Are the 50 Most Promising Startups You’ve Never Heard Of”(congratulations), so needless to say that would have also meant that the automotive industry was at your door. So?A:*Thanks, it was great to be listed in the Bloomberg article… I actually didn’t know about it until it was published!We’ve certainly been approached by automotive manufacturers and OEMs, especially since my background includes a lot of robotics and autonomous vehicle research. In fact, as an academic researcher, I’ve also worked on projects designing sensors and software tools for vehicle dynamics analysis for automotive manufacturers. This puts me in a unique position to understand the field very well.Firstly, as a founder * CEO of a technology startup, one of the things I’ve learnt is the need to remain focused. Even small distractions or changes in direction can consume resources unnecessarily and can have an impact on your team. Staying focused means that your team isn’t spread thin and you can continue to service and delight your current customers and maintain quality work. In addition to all that, some of the very largest companies in the world (which includes automotive companies) can be a large resource drain on a startup. You always need to be mindful of this as a leader of a startup.Secondly, as an engineer, there are some significant differences between a vehicular platform and a smartphone-based platform. As Navisens provides a Software Development Kit (SDK) for smartphones, we avoid custom operating systems and an operating system which isn’t iOS or Android. This rules out certain vehicle integrations. Since we’re tracking smartphones which are carried by humans, we don’t have the benefit of a comparatively much more stable platform such as a vehicle, and we don’t have the benefit of knowing the control inputs (e.g. steering, brake, throttle) of a human. Thus, we’re solving an extremely challenging problem, one which requires us to have an extreme focus on technology.Considering the above, I will mention we are actively in discussions with one or more automotive companies in some capacity