Bridging Academia and Innovation, with Mohammad Mahoor

In this podcast episode of the Entrepreneurship at DU podcast, Joshua Ross interviews Mohammad Mahoor, a professor at DU’s Ritchie School of Engineering and Computer Science. Mahoor discusses his transition from academic research to entrepreneurship, particularly focusing on his work in artificial intelligence and robotics to create social robots. These robots, such as Ryan, are designed to assist older adults with dementia and children with special needs. He emphasizes the challenges and rewards of balancing multiple roles, from academia to business, and shares insights on how interdisciplinary collaboration and AI advancements are driving innovation in his field.  

This episode is perfect for aspiring entrepreneurs, faculty members interested in commercializing research, and anyone curious about the intersection of AI, robotics, and healthcare.

 Transcript:

Joshua Ross:

My name is Joshua Ross and welcome to the entrepreneurship at DU podcast.

Mohammad Mahoor:

I’m a person that wanted to create things, build things, right? I’m an engineer and I don’t want to just do theoretical research or research just to, for the sake of having publications, I want to have an impact on people’s life.

Joshua Ross:

On this episode, I sit down with Mohammad Mahoor, a professor at Du Richie School of Engineering and Computer Science moho, an innovative researcher and a successful entrepreneur, discusses how he navigates the intersection of academia and business and how he’s developed the tools to build a thriving company while balancing multiple professional roles.

Mohammad Mahoor:

But having a product right understands customers need going out and talking to them and be able to translate their needs and what they want into a product that would risk the gaps. That’s a different story.

Joshua Ross:

Mohamed Maur groundbreaking work in artificial intelligence and robotics addresses the critical issues of dementia and loneliness among older adults.

Mohammad Mahoor:

So we decided to focus on senior care facilities, having a social robot that can help people, older adults with dementia, or those who are living lonely in senior care facilities.

Joshua Ross:

Here is my interview with Mohamed moho. Mohamed moho. Welcome to the podcast. I’ve really been looking forward to our conversation In the past. We’ve had entrepreneur alums on the podcast. We’ve had DU student entrepreneurs on the podcast, but you are our first faculty entrepreneur on the podcast. So welcome.

Mohammad Mahoor:

Thank you, Joshua, for having me.

Joshua Ross:

So you’ve been at the University of Denver for 16 years. You started as an assistant professor, and now you are a full-time professor since around 2018. So I’d like to get a little bit of background. So how did you end up at DU? Give us the pre DDU history and your background.

Mohammad Mahoor:

Yes, so I went to the University of Miami to get a PhD. I was so obsessed with image processing, computer vision back then, and I got my research was on face recognition. That’s what I did for my PhD. And then I was a postdoc in psychology department and I learned what we did for face recognition, face identification, to do research on kids with autism and to mostly look at the facial expressions of kids with autism. And after I did my postdoctoral training, I was looking for a job and I really love research and teaching and I was lucky that I was offered a position at the University of Denver as an assistant professor, although I wanted to join Microsoft and also some startup companies that they gave me an offer. But I think I was lucky that I could join du, and that’s how I continued to know my research on facial expression recognition. And then I got into robotics.

Joshua Ross:

Well, it’s interesting I’m doing the simple math on those that timeframe and Microsoft and some of those other technology companies, I guess were right down during that financial meltdown or a little bit before that.

Mohammad Mahoor:

Exactly, exactly. And the job market was really bad back then and it was really hard to find job. But I was lucky as I said that, to get a position here at du.

Joshua Ross:

Well, I think DU is lucky to have you as well. Thanks. So you have this PhD and you have all this academic research that you worked on at the University of Miami around facial recognition. How do you develop practical applications and turn it now into the social robots that we’re going to dive into today?

Mohammad Mahoor:

So that’s another good question. So it really takes a lot of effort research and also you need to know the gap between your customers and what you’re doing, your iCal research, and also working with an interdisciplinary team that would help to shape the research and also the connection between the applications and the theoretical research and to be able to come up with a product that would address the needs of customers.

Joshua Ross:

So to provide context for our listeners to kind of level up, what is a social robot?

Mohammad Mahoor:

So a social robot is a robot that is designed through support and assist people. We at the social emotional level, and instead of physical interaction, physical support and assistance, it provides social support, emotional support. So you communicate and engage with a human, especially people with social needs. For example, kids with autism, older adults with social isolation. So a social robot can provide those kind of supports.

Joshua Ross:

So why did you decide to focus on children with special needs and the elderly? Why those two kind of target markets?

Mohammad Mahoor:

That’s a good question. So my research before joining DU was on emotion recognition technology, how we can, for example, recognize facial expressions. And when I work with children with autism, I realize that technology can assist them, can help them a lot in many ways because they may have issues with interacting with human beings, but when it comes to technology, when it comes to toys, robots, they feel more comfortable. And that was one side of the story. And then after I joined DU, I realized that robotics, it’s really notable here, I mean that I found a few faculty members who do research on robotics. And then I thought that we can combine the two. We can use social robots as assistive technology or assistive tools to help children with autism. That was the beginning of my career and my research in social robotics. Then we looked around and see what else we can do. The first project I believe we had with social robots for older adults was to create social robots that can help people with depression, late life depression, older adults. That was another project. A group of faculty applied for funding and we got a grant from the National Science Foundations. And that’s how we started creating our first robot. That was Ebert, the expressive bear. That still is in my lab and that’s how we, my research actually, social robotics actually started.

Joshua Ross:

So are you saying that your research helped drive these two different target markets? Was it because of the need for this social interaction, but there’s such a large need of the number of the elderly population that we see growing and the number of these children with special needs?

Mohammad Mahoor:

Actually, it’s both the number of, for example, children with autism or older adults with dementia or depression, and also the lack of caregivers, care providers that can help or therapies. When it comes to depression, that’s all of them help. Or was the motivation or the motivation to provide solution like social robotics to address the needs?

Joshua Ross:

So have you always been an entrepreneur? I listen to different podcasts and I hear interviews and I read articles about different entrepreneurs and Mark Cuban was selling trash bags door to door when he was a kid and it subsequently had other types of businesses. Was that who you were or did something around your background, electrical and computer engineer shape your approach to entrepreneurship and innovation?

Mohammad Mahoor:

I always wanted to be an entrepreneur before doing PhD. I started an MBA 20 years ago, but then I dropped out, I came to the United States to do PhD and also I worked in a small business that I was an IT manager and I loved working with engineers and developed tools and software programs and so on and so forth. So it was always in the back of my mind to become an entrepreneur or have a business. I love that. And also I’m a person that wanted to create things, build things. I’m an engineer and I don’t want to just do theoretical research or research just for the sake of having publications. I want to have an impact on people’s life. And so that’s me and that’s why I got into this and I love what I’m doing, although it can be hell from time to time.

Joshua Ross:

I love that because I think research is very important, but I always judge research by the impact it has, the purpose it provides, and at the end is does it actually move humanity forward? So I love the fact that you’re saying you love the research, you love what you do, but you want it to be impactful.

Mohammad Mahoor:

Oh, absolutely. I mean that it’ll definitely move humanity forward and that’s the purpose of social robots that we have been working on to improve the quality of life of people. And I think that’s the impact. That’s what I have seen so far and more to come.

Joshua Ross:

Alright. So your company is Dream Face Technologies, right? Started around 2014, so roughly 10 years ago. So when you started, what was the initial focus? Was it just you, did you have a partner? Did you have to build a team? Walk us through the start of Dream Face.

Mohammad Mahoor:

So right at the end of my first NSF grant, that was a collaborative grant that we created and also we had the idea of a rear projection robot then do you file for a patent to protect the invention that we had? And at the same time I applied for another NSF grant that was icorp. It’s like a bootcamp that teach you how to have a startup company, what are the elements of having a startup company and how to do customer discovery, customer identification, business model development, and so on, so forth. So I went through that icorp program with one of my PhD students and also I had a mentor from industry. And so that was really fascinating. That was a great program and I learned a lot on how to have a startup company, what it takes to have a startup company. Customer discovery is super important.

So after we finished the Icorp program, that was a two month very intensive program and at the end of the day we tested our hypothesis whether having a social robot can, as a product, can build a business based on that. And back then we had seven different customer segments in mind. We tested our hypothesis that which one can be the right target market. So for example, a social robot can be used as a concierge, as a receptionist in hotel, can help children with autism, people with a special needs, older adults with depression, dementia, those who are lonely and so on, so forth. And at the end of the day, after doing customer discovery, after interviewing more than a hundred potential customers, stakeholders, and also going to the building or developing a business model, we decided to have a startup company. That’s what it was, a go or no go decision at the end of the bootcamp. And that’s how I founded Dream Face Technologies back in 2014.

Joshua Ross:

Okay, so your customer segments then at that point in time or what?

Mohammad Mahoor:

So we decided to focus on senior care facilities, having a social robot that can help people, older adults with dementia, early dementia, or those who are living lonely in senior care facilities because of lack of caregivers, shortage of caregivers. So social robots can fill the gap as a tool, assistive tool to help the owners of senior care facilities and as end users, older adults, residents of those facilities.

Joshua Ross:

So you’re talking about Ryan, correct?

Mohammad Mahoor:

Yes. I’m talking about Ryan.

Joshua Ross:

Right. So this is a great segue. So introduce us to Ryan. How does it work? How does it assist these early onset, I believe you said patients?

Mohammad Mahoor:

So Ryan is a social robot that has several, is a humanoid robot first of all. And the face of Ryan has created based on rear projection technology, meaning that with computer graphics we created different faces and we project the face onto a translucent mask. And the advantage of these kind of technologies that you can change the face, the appearance of the face, and also we can have a very accurate visualist speech. The speech is synchronized with lips movement and that helps when it to communication and conversation with the robot and you can change the appearance so you are not stuck with just one design for the face of the robot. So that’s how we create, that’s the technology behind this robot. Then we added a lot of AI models to Ryan and you can have natural conversation conversation with Ryan. Ryan can recognize your facial expression so Ryan can empathize with users, understands their feeling, and for example, if they are not happy, Ryan asked what makes you to not to be happy?

So empathy is another aspect of Ryan. And then we added several games, brain games to Ryan so you can play games, these games with Ryan, either like Ryan to be your opponent or Ryan to be your friend that you play together. And also we added some other modules. All these modules and all these features that we added to Ryan was based on the research that we did. We created the first prototype of Ryan in 2016, and then we did research him and robot interaction studies and we got feedback from potential users and then we revise the features. And then also at the same time we work on the aesthetic of Ryan, the hardware and improve the features of Ryan. We have done a lot of modifications to come to this version that we have. And as I said, it’s been designed to assist people with earlier stage dementia and social isolation and depression because it would engage them in social conversations and also in physical activities. For example, Ryan is a yoga seated yoga instructor and can instruct people to be physically active and do yoga with Ryan. So it’s a collection of or a set of features that are essential for people who have earlier stage dementia, social interaction conversations, physical activities, brain games, even music.

Joshua Ross:

Can it work both one-to-one with the human or also one to many? Because I’m guessing with the yoga and some of these games it can work with a wider audience.

Mohammad Mahoor:

Yes. So it’s initially designed for one-on-one interaction. And when it comes to conversation still, Ryan is not capable of answering multiple people’s question, but when it comes to yoga for example, or some of the physical games that we have, it can be one too many interaction.

Joshua Ross:

So what’s the current status of Ryan? Is it available for purchase? Are you still beta testing it?

Mohammad Mahoor:

We have finished testing, Ryan in the past five, six years. We have tested Ryan with more than 50 residents of six different senior care facilities, including assisted living facilities, independent and memory care units in the Denver Metro. And we kind of finalize the features of Ryan, although when it comes to ai, every day something’s new is coming up and we need to improve our models. But mostly we are done with testing and in this stage we are looking for early adopters to lease Ryan to rent Ryan and give it to their residents. And also at the same time you’re looking for investors to be able to scale up the business and work on mass manufacturing of Ryan. So we’re in the stage of pre-launch, pre-marketing and looking for early adopters.

Joshua Ross:

So these early adopters would be assisted living care

Mohammad Mahoor:

Facilities? Yes. Yes, absolutely.

Joshua Ross:

Okay. You’ve transitioned from academic research to entrepreneurship, which we’ve established with Dream Face Technologies and you’ve had some success with that. Walk us through some of the biggest challenges as you’ve done from taking this research to commercialization.

Mohammad Mahoor:

That’s a good question. So that’s really a big challenge. It requires a different skillsets and mindsets and being an entrepreneur can be held, your life can be held. It’s not something that you have to be prepared to face. Challenges. Fundraising is one challenge. Building a team is another challenge. Finding talent with the right talent that would work with you and delegating the task to them. And also day business operation, financial management, budgeting, even sometimes you have to sit down with your team and when it comes to coding or research or building a robot, that requires a lot of manufacturing and machining and so on and so forth. So it requires a person that is versatile and also is able to pivot and also put different hat on every time based on the needs and the challenges that the company has. But I would say having a good business partner would help a lot. Someone that understands business and marketing and is less involved in research and technical development. I think it’s better to decouple these two to some extent. But although the business person should understand technology to some extent and vice versa, but it’s better to a separate these two tasks.

Joshua Ross:

I mean, that’s what makes a great partnership is partners with the same values and the same work ethic, but very different skill sets. Absolutely understanding what each others do and respect that, but can go off and do things that the other partner can’t do. Otherwise you have to go out and hire that person.

Mohammad Mahoor:

So it’s better to have identify that partner from day one and do it together other than getting to a point and then finding a business partner. So you want someone who has a skin in the game from day one early on, other than being someone on board to help you. Although if that’s possible, I’ve seen some businesses who have done that, but I think the chance of success is higher if you have a business partner that can compliment you along the way.

Joshua Ross:

Yeah, that’s very true because also being an entrepreneur can be incredibly lonely and having someone to talk to about it and empathize and absolutely share the ups and downs. I completely understand that. You talked about wearing multiple hats and then you were talking about it within the business, but you also do it as a professor, researcher, entrepreneurship innovator. How do you balance all these roles because you can’t clone yourself yet?

Mohammad Mahoor:

Absolutely. That’s a very good question. From time to time, either I took the leave of absence or I was in sabbatical, right? Two times to be able to work full-time or part-time, more than 51% of my time is spent in my company. It’s very challenging. It’s really challenging and requires a lot of dedication and time commitment. I would not recommend it to everyone to do that. What I did, if time goes back, I learned a lot. And if time goes back, I would eventually do it slightly differently, but I would be happy to share my experiences with faculty who wanted to do this. And I have some recommendations, advices. And indeed, I gave a presentation a couple of years ago to, there was an NSF workshop meeting as a faculty who had to have a startup company. I talked about the challenges, the issues that you may face and when to do it, when not to do it. Based on my experience, I’m happy to give a presentation and talk about that or at least mention a few the challenges.

Joshua Ross:

I think that’s a great idea and I’m going to put a pin in that one and reach out to you on that because I think a lot of faculty, not only here at the University of Denver, but around the state of Colorado would get a lot of value out of it, especially where you are with your company and what you’ve gone through. So this next question, I’ve been really excited about asking more for your answer because it’s something that I’ve been very curious about. So you live in this academia world world, but you also had to go to industry to develop Ryan and get the product out and both of your social robots out into the world. You also work with these not-for-profit organizations, for funding and for grants as well. How does a collaboration work amongst industry academia and has it helped you? And also what have been some of the challenges?

Mohammad Mahoor:

Yeah, so first of all, the research that I’m doing is interdisciplinary. So I’ve worked is a team effort. I have engineers on team researchers. I’ve worked with psychologists, caregivers, nurses, geriatricians and so forth. So it is a team effort. So when it comes to research and creating a product, but then when you go out and talk to investors and VCs and also different industries to, for example, purchase hardware or pieces that you need to build your robot, that requires a different skillset and mindset and it’s a learning curve. It takes some time to learn and also learn from other people’s mistakes and also my own mistakes that how to do it.

And also I learned from one of the former faculty that we had here where the startup company and the company kind of failed and they had to dissolve it. And I learned from that faculty as well. So it takes time and effort and you got to be willing to learn and pivot, as I said, and go to a lot of meetings, be open to ideas. And another thing that have helped me recently meeting you and also becoming familiar with your program, the entrepreneurship program at du and also Daniels College of Business. And I started a program or enrolled in a certificate, actually program to have a certificate in business foundations hopefully in a year. And that helped me a lot to understand how to run business and what it means to run a business because that’s totally different from running a research lab or writing grants because in academia or as a research faculty or faculty who’s doing research, mostly all the care is to have successful research and have publications at the end or maybe patents, and also advise our students to successfully defend their PhD dissertations or master’s thesis. But that’s it. But having a product, right, understanding customers need going out and talking to them and be able to translate their needs and what they want into a product that would address the gaps, that’s a different story.

Joshua Ross:

So how does the University of Denver support your commercialization of research?

Mohammad Mahoor:

I think University of Denver has been very supportive so far. I mean the office of technology transfer device provost for research, and also my dean, the previous deans, our department chairs, and also students. I’ve been lucky to be able to find talents and good number of students at DU to work on Ryan. I can say that Ryan is a product of DU because more than 90% of people who have worked on developing Ryan are DU graduates. So I think we should be proud of our engineering and computer science program, both, and also then the mechanical engineering department and the facilities that we have and the openness and the generosity of some of the deans that we have had here to support me, to help me to allow me to use resources at du. And also from time to time, Daniel’s College of Business has been great research for me to be able to go and talk to people, give me advices now the program that I’m enrolled in to hone my skills in business development.

Joshua Ross:

Great. And we’re a big fan of the Office of Technology Transfer. We had Art Terelli and Ashley Thomas on the podcast a couple months ago, and they walked our audience through the whole commercialization process and how it works. So a few questions to close out our conversation. What is your vision for the future of human robot interaction in both healthcare and education? And how are you positioning yourself to work to address these future needs? What are you doing today? But as you see the rapid evolution of technology, or not evolution, but how quickly it’s evolving? Yeah,

Mohammad Mahoor:

The technology is evolving really fast, and I envision or foresee that in the near future, maybe in 10 years or even less than 10 years, five years, you will see social robots that are integrated into our daily life activities, although there’s still some stigma and pushbacks from some folks, but I see that is coming and it’s happening. AI helps a lot to be able to have more sophisticated, actually social robots that can have better, more engaging interaction with people, especially the large language models that were recently development release, going to help a lot with the future of social robots, human robot interactions. And they can help in a lot of areas and fill the gaps, for example, lack of caregivers, workers, laborers and so forth. And also, it’s going to help us when it comes to some tasks that are repetitive or some tasks that are really hard and challenging to do, robots can do them better, more accurately and precisely.

And also some of these jobs might be very dangerous. And so we can pass those jobs to robots, but then we focus on other things. I believe that very soon, as I said, less than 10 years, you’ll see social robots that are really integrated into our life. One thing that might be cost prohibitive for now, it’s the cost of robots to build robots and hardware. And I hope that we can reduce the cost and make it more affordable for people or come up with different business models, for example, subscriptions or renting a robot that you can rent a robot instead of purchasing a robot and make it more accessible to more people.

Joshua Ross:

So you’ve been using artificial intelligence for a long time in your robots, and 18 months ago chat, GPT was released to the world, and we’ve seen a rapid adoption in both people using generative AI and the number of companies that have come out with their large language models. Where do you see this advancement in these tools helping your robots out?

Mohammad Mahoor:

So Ryan has more than six AI models or uses six AI models for different tasks. In fact, without ai, Ryan wouldn’t be able to function at all. So for example, we use AI models for face tracking, for emotion recognition, for voice synthesis, for voice recognition, and also for conversation. And even some of the games that we have, they use AI models as well. So we have been using, even before the release of chat g pt, we use a large language model from meta that is integrated into Ryan for conversation with older adults. But these new models that are larger, more sophisticated, and they have more parameters that help to have more engaging conversations with users. And also these models are going to be mostly multimodal. The new coming models, for example, GPT-4 oh, which always stands for Omni. It’s a vision and language model. We call it foundation models. And I believe that these models are going to even help Ryan to be better in a sense that Ryan, for example, understands the environment, understands where Ryan is, or where different objects are left or put. And then Ryan can remind people about where they leave or what they have done in the past. So I believe significantly these models, AI models can help to advance Ryan and social robotics and human robot interaction in general.

Joshua Ross:

Does Ryan get smarter more Ryan interacts with humans, not just through supervised learning, but actually just one-to-one interaction with, let’s say a patient?

Mohammad Mahoor:

Yeah, so Ryan has the memory, maybe it’s a short-term memory, but we can set a parameter to say that you can learn the past five conversations or remember, or 10 past conversations. So it does kind of have a memory. And also Ryan is personalized, meaning that every individual has an account, an ID and a card. So when you log into Ryan, then Ryan knows that that’s you and the conversation. It’s kind of personalized. And also is your robot, Ryan knows your name, your favorite topics, even songs and so on and so forth. So to that extent, yes, Ryan learns and memorizes the past interactions. And that would help when it comes to a more engaging and national communication conversations with the robot.

Joshua Ross:

And I would have to assume that’s only going to get better over time as Ryan learns gets to know the person with whom they’re interacting with.

Mohammad Mahoor:

Yes, absolutely. And we think we are going to add one of these recently released large language models to Ryan. And that would help. For example, we would like to add a cognitive behavioral therapy to Ryan that would help with anxiety and depression therapy. And I think a large language model that knows you in terms of your right, what makes you to be anxious or have depression and what is going on in your life, that would help a lot.

Joshua Ross:

So my final question for you today is what advice do you offer young researchers, think back to your time at Miami as you’re looking at taking this research and looking at commercial opportunities, how to translate that work into innovative real world applications?

Mohammad Mahoor:

So my advice would be to, I mean that don’t limit yourself to what you’re doing and give it a chance. Go and explore. Maybe take some courses, Daniel’s, red College of Business, or maybe enroll in your entrepreneurship program or book camps, the plug. Yeah. And so to be curious and read maybe one or two books, listen to podcasts and see what it takes to be an entrepreneur and whether are the pieces and the elements and that are essential for being an entrepreneur. And of course you got to, to be open to learn new things and go out of your comfort zone. As I said, it can be, your life can be hell, right? And you have to deal with so many challenges and be busy all the time. It is not like a nine to five job. It’s not, it’s something that it keeps you busy 24 7. But one thing that I have learned all these years is that you have to better, or you should have delegates that you can pass on some of the task that you may not be good at it to others and focus, but know what’s going on. You have and what it takes to do those tasks. But you may not be perfect to do those tasks. And also building a team is really super important, and finding the right talents I think is important.

Joshua Ross:

That’s great advice. Well, Mohamed, thank you for being on the podcast. Thank you for your time today, and thank you for making me a little bit smarter.

Mohammad Mahoor:

Thank you, Joshua. Really appreciate it. Thanks for having me.

Joshua Ross:

The entrepreneurship at DU podcast was recorded in Marjorie Reed Hall on the University of Denver campus. You can find us on Instagram at du Entrepreneur on Twitter, X at DU entrepreneur, and on Facebook at entrepreneurship at du. This episode was engineered, edited, and produced by Sophia Holt. Entrepreneurship at DU is part of the Daniels College of Business, which has its own podcast. Check out Voices of Experience wherever you get your podcasts.