Operating Robots with VR with Dr. Chang Liu, CEO & Chief Designer at Extend Robotics

Paul David (00:05.347)
Hello everybody. Welcome to a new episode of the Tech Can’t Save Us podcast brought to you by Literal Humans. My name is Paul David, I’m the CEO of Literal Humans and one of your hosts. We’ve made a bit of a change here at Literal Humans, and I’ll be doing the hosting. Unfortunately, Maya, our former co-host, got poached by Sifted, one of the early-stage tech publications, so she’s left us, leaving me to manage the podcast. But the good news is we still have fantastic guests, like today’s guest, Chang Liu, the CEO and chief designer at Extend Robotics. We’re going to talk with Chang today about operating robots with VR. It’s an exciting topic—VR is hot, robotics is hot. So, welcome to the podcast, Chang.

Chang Liu (00:52.364)
Yeah, thank you for the invitation, Paul. It’s great to be here.

Paul David (00:56.262)
Yeah, so just by way of introduction, we want to welcome you and mention that you’re the founder and CEO, and also the chief designer at Extend Robotics. What Extend Robotics does is use consumer VR tech to make it easier for people to control robotics remotely. Rather than needing to be right next to a robot, which is what we usually see, sometimes people are working in potentially hazardous environments, and working from a distance increases safety. It’s a huge evolution in robotics technology. This approach allows operators to control robotic arms more intuitively, as if they were their own limbs, which is massive. Using VR interfaces enables more natural control, allowing more complex tasks to be completed remotely. So, by using consumer technology, Extend Robotics aims to increase the number of people who can operate robotics, as the name suggests. Essentially, anyone with minimal training should be able to tele-operate and automate robotics. This caught the eye of Peter Jones at BBC’s Dragon’s Den, after you used it to pour a beer—a very famous clip. After the Dragon’s Den appearance, you all got some investment. We won’t mention how much. I wanted to ask, Dragon’s Den is very popular. How did you decide to go on the show?

Chang Liu (02:33.89)
Yeah, we were approached by the BBC, and at that point, we were quite young and looking for more exposure. Pouring a pint of beer seemed like a perfect demo for a UK audience—it’s very topical, and everybody is interested in how it’s done. I think it was a great opportunity to raise awareness for a new technology, especially by showing it working in real life in front of the public. It really made a difference because people still think it’s something out of sci-fi and not real. But here it is, already working. It played a big role in educating people about how far our technology has come.

Paul David (03:35.61)
Yeah, you picked a great venue—pouring a beer. You clearly know your audience well. That speaks volumes about your growth strategy and your understanding of your audience, which we’ll get to a bit later. But I wanted to take a step back and understand what inspired you to found Extend Robotics in the first place.

Chang Liu (03:54.102)
It’s a good question. We are a human-centric robotic company. We want to make sure robots aren’t turned into terminators. We imagine a future where we live cohesively with robots, and ultimately, they extend human capabilities and empower people to do their work better or live a better life. We also want to solve real-world challenges. For robotics, it’s mainly about amplifying or extending labor productivity in the real world. There’s a tremendous labor shortage coming in the 21st century. For example, in manufacturing alone, we’re forecasting about 2.4 million jobs unfilled by 2028, leading to about an 11 trillion GDP gap. That’s a huge problem, probably the biggest problem in the 21st century. There are already many robots in factories, but many critical tasks still require people to work in harsh or hazardous environments. Traditional automation struggles with the complexity required for some tasks. Many jobs that need flexibility just can’t be automated. To address the labor challenges in the real world, we aim to solve this shortage. We’ve built embodied AI robotic solutions to provide a new approach that ensures safety and reliability in tasks that couldn’t be automated before.

Chang Liu (06:19.491)
We’re trying to empower people to do their work in a much safer and flexible way. That’s what we’re trying to do.

Paul David (06:26.992)
Yeah, it’s interesting. There are so many layers to it. I think about this when I’m cycling around the city—people are doing really dangerous, dirty jobs. There’s definitely an element of leveraging robotics to replace those humans and hopefully retrain them to do safer, more interesting work. Then there’s the issue of just having a shortage of workers. You’re addressing both: sometimes replacing people for safety, and sometimes augmenting them. Just for context, the manufacturing sector is going through a massive labor shortage right now, with an estimated 2.4 million jobs unfilled by 2028, which will lead to a potential $11 trillion gap in GDP. As Chang mentioned, there are loads of critical tasks that still require human intervention, so it’s great to see companies like yours coming up with solutions that address the labor shortage, safety, and security for workers.

Chang Liu (07:38.626)
Exactly. There’s a big challenge ahead, and we’re here to help.

Paul David (07:44.678)
Yeah. Talk to us a bit more about your ambitions for the company. Who do you want your main buyers, customers, and users to be? Where do you want to take the company over the next three to five years?

Chang Liu (08:01.41)
To address the labor challenges, we need to scale automation quickly. We want to help the world transition from traditional rigid factory automation to real-world, flexible, intelligent automation. Our approach is to embody human-like intelligence—such as chat GPT or more recently, Llama or DeepSeq—into a physical robot to enable AI capabilities similar to what a person can do. But here’s the catch: training a robot is much harder than training something like chat GPT because chat GPT already has vast amounts of data from the internet, while robots don’t have that kind of data. It’s not an overnight miracle to train an AI model for robots. We need to bootstrap the foundational AIs by leveraging all possible sources of data and iteratively collect more data as we use the robots. The key is to deploy robots and make them useful from day one to encourage adoption. Then, once they’re in use, we can collect enough data for the AI to improve. This feedback loop is critical. It’s called a data flywheel, and unlocking it is the holy grail of AI. Our ultimate goal is to build automation that can perform any task like a human, using software that can be deployed in various applications. For example, in nuclear decommissioning, workers are exposed to radiation and can’t work for more than two hours a day for their health. We’re also exploring space manufacturing and harvesting grapes for UK vineyards, where labor shortages are a major issue.

Paul David (11:05.936)
Hmm.

Chang Liu (11:23.79)
We’re focusing on manufacturing applications, particularly where it’s dangerous for humans to work. We’re addressing needs in industries like electric car battery production and car painting, where labor shortages and automation needs are severe.

Paul David (12:00.87)
Hmm.

Paul David (12:07.558)
You mentioned a lot of needs, and that’s a great segue into my next question. We’re a growth marketing agency, and we help clients develop and execute their growth strategies. We focus on branding, content marketing, SEO, paid social, PR, and more. So, we have to ask you: how have you grown so far? You mentioned getting some funding. How do you plan to grow over the next year or two? What strategies are you putting in place? Who are your buyers, and how are you communicating with them throughout the sales cycle?

Chang Liu (12:47.562)
As a company, we focus on software. We developed a VR application called AMAS, which has been deployed for the last three years. It’s an intuitive application to gamify robotic operations for a wide range of third-party commercial robots, including robot arms, end effectors, cameras, and mobile bases. The app serves three main purposes: first, to operate robots remotely in real time (teleoperation); second, to deploy and supervise imperfect AI systems in robots, with human involvement to mitigate failures; and third, to collect data for training AI models continuously. It’s already being used in many applications, and we’re revenue-generating. A lot of customers are coming to us, and we’re helping them build suitable solutions. We’ve historically deployed across various sectors, including nuclear and space, but going forward, we’re focusing on manufacturing applications due to labor shortages and cost reductions.

Paul David (15:03.206)
That’s really interesting. Can you tell me more about the buying cycle? How does someone engage with you and purchase your solution?

Chang Liu (15:16.684)
We’ve built a strong online presence, so many customers find us through the web, submit a form, and come to us with challenges. We post videos and content on social media and YouTube to show people that the technology is real and working right now. We want to prove that this isn’t just sci-fi—it’s available today. We also attend events and exhibitions to show the technology in action. A lot of it is about educating people and building their confidence that they can use it. They don’t need a PhD—just the ability to play VR games. We aim to make it easier for people to adopt the technology. We also partner with robot hardware vendors. For example, if customers already have robots, they can simply add our software as a new feature without changing their existing hardware.

Paul David (17:41.666)
Hmm.

Paul David (18:06.406)
Yeah, interesting. I like the convenience of meeting people where they are and letting them use their existing robots. It’s about understanding the customer’s specific needs. That’s huge for us—our team is trained in behavioral science, and we work with a partner agency that has a framework for agencies. It’s important to embed human-centric thinking into everything we do. It’s about engaging with customers on their level and helping them move forward. Whether through robotics or marketing, we need to have empathy and an understanding of their needs at the ground level.

Chang Liu (18:54.102)
Exactly. Also, many people worry that AI robotics are going to take their jobs. I think very long term, that’s probably the direction we’re traveling, where ultimately you no longer have to work, but you can still work if you prefer that particular lifestyle. But in the short term, it’s more about empowering people. You’re not replaced by AI robotics, but you may be replaced by people who know how to use it and who are open enough to start using it to empower their business. So yeah, we really see this as empowering tools that people can use, rather than their competitors.

Paul David (19:42.502)
That’s great. Would love to ask you about your role in sort of the clean future, you know, so clean tech. You all participated in the Connected Places Catapult Clean Future Program. It’s a mouthful. You know, tell us what this has involved and how do you see Extend Robotics contributing to a clean future?

Chang Liu (20:02.966)
Yeah, absolutely. We’re currently, we’re really lucky to be part of this Clean Future program initiated by Connected Places Catapult, which gives us a lot of resources to explore applications, particularly in the electric automotive manufacturing industry, especially building electric cars. And that, I believe, is the major driving force towards a cleaner future. Yeah, our role there is looking at the objective to improve the advanced production and recycling process of electric cars, particularly focusing on supply chains, using robotics and AI. We’re particularly focused on reducing costs and increasing throughput by introducing new ways of automating tasks. Because traditionally, even though the automotive industry is the most automated industry in the world, there’s still a tremendous amount of work being done by people. These challenges are still present because traditional automation hasn’t been able to address them fully. Our role is to focus on these difficult cases and see how we can expand and deploy this new AI or human-in-the-loop framework to help with tasks that couldn’t be automated before. So, yeah, we’ve been involved with the Institute for Advanced Manufacturing and Engineering at Coventry University, and they’ve provided a test bed for us to deploy and use our robots in. So we’re looking at testing the recycling of car parts and also building and assembling batteries for cars, using our robots. The idea is that you can teleoperate them in the beginning, then use the data to automate the whole process. Obviously, it’s very hard to automate battery assembly tasks, so that’s just one of the examples from the automotive industry, but it’s a good starting point with the resources secured for that. Thanks to the program, we’re already in discussions with a few partners in the electric car supply chain, such as UniPark and Hyperbad.

Paul David (22:59.334)
Thanks for that.

Chang Liu (23:00.02)
Specifically on those battery productions. I’m really looking forward to further development. We have some videos on our YouTube channel showing more details of the trials and testing, so feel free to check those out.

Paul David (23:20.89)
That’s really cool. Yeah, it’s interesting that these industries, which are typically known as being “dirty” or not obviously associated with clean tech, also present a huge opportunity to clean them up, green them, and innovate. That’s really cool, getting into the green transition, clean tech space with what you’re doing. And thank you for that work. It’s much needed. The planet thanks you.

Chang Liu (23:52.736)
Yeah, we’re just getting started. There’s an exciting future ahead.

Paul David (23:57.894)
Yeah, that’s really cool. Well, awesome. Wanted to jump to our quickfire questions. These are just kind of fun, personal questions for the audience to get to know you a bit more as a human. And I think you’ve already thrown in some great answers here. But yeah, I want you to give a little bit of voiceover. And again, they’re quickfire, so just say your initial response—what comes to mind. But the first one: If you had to found another tech startup after Extend, what would it be and why?

Chang Liu (24:28.718)
Yeah, there are quite a few fields. I think in the next few years, it’s really just about how you apply AI to all sorts of applications. AI in general is a tremendous invention, similar to the invention of electricity or the internet. It enables so many different applications. We’re just getting started with AI, and there are so many applications it could be used for. Like gene editing or even just discovering how genes work. Could we even use genes for therapy and improve healthcare? Another idea is using AI to decode brain information and create a brain-computer interface. It’s interesting how progress is being made in this field. You can also use AI for education. I think it’s great to have a personal tutor that’s as smart as Einstein, and you can simply ask them questions. They’ll give you the perfect answer and guide you based on your learning style. This would dramatically increase educational efficiency and quality. I see a large opportunity in that area. Yeah, if I had some more free time, I’d look into those.

Paul David (26:22.298)
Yeah, that’s the fun stuff—how much these new evolutions of tech are going to give us more free time and help us do the stuff that we don’t want to do. That’s the interesting part. Along those lines, what tech invention makes you the most excited and why?

Chang Liu (26:32.92)
Yeah, from what I’ve observed, I think people are using humanoid robots. There are so many reports showing that humanoid robots are being built and deployed in factories, walking around, running, and doing useful tasks. I think it’s exciting because this is such a generic form factor for humanoid robots to be built and start being useful. I think it’s a great turning point, where you can deploy AI in humanoid robots, and those robots will do things that a person can do and be as clever as a person. It could become your colleague, your friend, or even a trusted family member. I think that combination of humanoid robots and AI will be revolutionary.

Paul David (28:06.278)
Yeah, it’s really interesting. Have you seen the film After Yang? I think it’s called that, with Colin Farrell. Have you seen it? It’s about a family that has a humanoid robot, and yeah, I think it captures what you’re saying about the potential for bonding. A lot of people out there are lonely, don’t have companionship, or someone to talk to. Obviously, it’s no full replacement for humans, but it can fill a gap potentially. I think that’s really interesting. I agree with you, it’s exciting.

Chang Liu (28:12.846)
Right.

Chang Liu (28:16.366)
Yeah, exactly.

Paul David (28:36.232)
Thank you.

Chang Liu (28:37.578)
Yeah, it becomes even a philosophical question: What is a person? What is a being? You can certainly connect emotionally with robots, and it becomes quite interesting.

Paul David (28:44.634)
Yeah, yeah.

Paul David (28:55.174)
Mm-hmm.

Paul David (29:13.91)
Yeah, the philosophical implications are fascinating. We’ll have to ask this requisite question. Conversely, what makes you most concerned? What tech innovations make you the most concerned at this current moment?

Chang Liu (29:13.91)
Yeah, as I mentioned, AI is becoming so powerful and fundamental to human society, and its safety is becoming concerning. We need to make sure that, as with any powerful tool, it doesn’t become a double-edged sword. We need to ensure that we’re not using it the wrong way, that we’re not developing it without consideration for safety or the potential hazards it could create. So, I think making sure AI is developed safely and aligned with human values is important. Transparency in AI development is critical. Not only that, but those who operate or supervise robots must have full control and visibility of what the robot is doing. If anything goes wrong, there needs to be an immediate way to intervene. As responsibility increases for robotics and AI, this has to be evaluated with a critical safety perspective.

Paul David (30:30.308)
Yeah.

Chang Liu (30:41.836)
Safety is essential.

Paul David (30:45.562)
Yeah, it makes sense. I mean, this is the answer we keep hearing when we ask this question—AI safety keeps coming up again and again. It makes total sense at this critical juncture. Switching over to talking about humans, we’d love to hear about a literal human in your life who’s been integral to your career journey.

Chang Liu (31:05.718)
Yeah, I think I would probably give Elon Musk credit for persuading me—well, planting the seed for me to pursue an entrepreneurial journey. His level of success and the kind of inspiration he provides, from technology to building it to benefit humanity, has really shaped how I approach life. His long-term vision for contributing to society really influenced my decision to follow this path.

Paul David (32:05.124)
Yeah. And did you have any personal interactions with him, or was it more from afar?

Chang Liu (32:08.87)
No, no. I was inspired by him, not necessarily linked to him personally.

Paul David (32:17.85)
Got it, got it. Yeah, fair enough. I didn’t know if you had any juicy Elon stories to share on the podcast, but that’s fine too. All good, all good. And just to round it out, we’d love to hear an inspirational quote. A lot of founders have inspirational quotes they kind of carry with them throughout their careers. Maybe one that stuck with you through your entrepreneurial journey in tech.

Chang Liu (32:24.461)
Haha.

Chang Liu (32:41.07)
Yeah, not necessarily inspirational. Well, I think, as Elon said, “Most of the time, if you need inspiration or encouragement to be an entrepreneur, don’t do it.” You don’t really need that. It has to be an internal passion for what you’re doing. But for something that helped me along the way, it’s a five-step process: How do you make a company work? This famous five-step process by Elon is quite helpful. You start by questioning the requirements, making the requirements less dumb, deleting unnecessary steps, simplifying processes, optimizing processes, then accelerating processes, and ultimately automating everything. This sequence is really inspiring and widely applicable. It’s useful in so many areas, including business or even personal life. I’ve found it very intriguing.

Paul David (34:33.026)
Yeah, we’ll have to share a link to that five-step process. I love the idea of simplifying things, especially in areas as complex as robotics, VR, and tech, helping people work through problems and understand things better. That’s a great one.

Chang Liu (34:47.31)
Yeah, yeah. The order is very important. You want to question the requirements first before deleting unnecessary parts. You want to automate before accelerating. If you get the order wrong, the process won’t work. So, it’s quite interesting.

Paul David (34:51.046)
Mm-hmm.

Paul David (34:56.953)
I know.

Paul David (35:14.166)
Well, that’s super interesting. This has been a fantastic conversation. Chang Liu, founder and CEO and chief designer at Extend Robotics, I’ve really enjoyed this conversation. Where can folks find out more about the great work you’re doing at Extend Robotics?

Chang Liu (35:30.242)
Yeah, the best place is our website www.extendrobotics.com, and we publish most of our stuff there. Also, check out our YouTube channel and LinkedIn.

Paul David (35:47.492)
Nice, we’ll share those links in the episode description. I just want to thank you so much, and to our listeners: Please give us a share, like, subscribe, five-star ratings so we can keep bringing on fantastic guests like Chang Liu, founder and CEO of Extend Robotics. Thanks so much for jumping on the pod today, Chang.