Interview with Under Control Robotics (UCR)
A Conversation on Humanoid Robotics and Entrepreneurship
In this edition of the MRSD newsletter, we are excited to share an interview with
Wenda Wang, an ex-ML engineer at Apple SPG, Founder Fellow at South Park Commons,
and an MRSD alumnus who is now the co-founder of Under Control Robotics (UCR),
a company focused on developing humanoid robots for heavy industries. From his journey
at CMU to his entrepreneurial endeavors, this interview sheds light on the motivations,
challenges, and vision behind UCR.
Let’s start by talking about your MRSD journey. How was the program for you, and how did it motivate you to pursue UCR?
Wenda: Thank you for having me! It’s hard to believe it’s been almost 10 years since I started MRSD. I joined in 2014, and at the time, self-driving cars were making waves, especially with Google X’s project. This sparked my passion for robotics and ultimately led me to apply to CMU. MRSD was an intense, rewarding experience, particularly in those first few months. The program was condensed into a year and a half for our cohort, which meant we were always busy—weekends dedicated to assignments, projects, and intense coursework. The program prepared me well for a career in robotics, offering a balance of theoretical and practical learning.
After graduation, you went on to Apple as a Machine Learning Engineer. What made you decide to transition into founding UCR?
Wenda: It was a great opportunity to work at Apple, especially during a time when autonomous systems were becoming more mainstream. However, after eight years at Apple, I felt the urge to build something from scratch. Silicon Valley’s entrepreneurial environment fueled this decision. Robotics was always my passion, and I wanted to work on something that could shape the future of automation, particularly in heavy industries. I had gained valuable skills at Apple, working on large-scale projects, and I felt it was the right time to move forward with my own startup.
What led you to choose humanoid robotics as the focus for UCR?
Wenda: It’s a mix of factors. There’s a natural correlation between breakthroughs in machine learning and advancements in robotics… With the rise of generative AI, I saw a similar wave coming for robotics, where general-purpose platforms could drive a wide range of tasks. Humanoids, as general-purpose robots, represent the next frontier. By focusing on humanoid robotics, I saw an opportunity to push the boundaries of what’s possible in automation, particularly in heavy industries.
Why did you decide to focus on heavy industries specifically for UCR’s robots?
Wenda: Heavy industries offered a more defined and less dynamic environment compared to other sectors. The need for humanoid robots in dangerous environments, where safety and reliability are paramount, was clear. Moreover, the economic value proposition is significant. Robots replacing humans in high-risk tasks could significantly reduce costs, such as insurance premiums and labor costs. Once we master humanoid robots for heavy industries, the technology could be scaled and applied to wider markets.
Can you describe the stage where UCR is now? Are you ready to ship your robots to customers?
Wenda: At this stage, we are focused on creating something that is truly useful. It’s really about building a product that people are willing to pay for. While the concept is simple, executing it well is challenging. Tactically, we need to be laser-focused on mastering at least a few tasks and ensuring high robustness and success rates in real-world deployments. Once we can perform at a high level in one or two key tasks, we’ll be able to expand our capabilities. It’s about delivering real value to our customers first before scaling.
Are you selling platforms and letting clients deploy their own algorithms, or is the software included?
Wenda: We’re still working on finalizing our approach, but we’re open to various collaborations. Essentially, we want our customers to use our hardware platform and be able to leverage different parts of our software and AI stack to optimize their applications. As a small company, flexibility and collaboration are key to how we work with our customers.
Does humanoid robotics have established safety regulations? How do you ensure your robots are safe for use?
Wenda: Humanoid robotics doesn’t yet have specific government-issued regulations like those for self-driving cars, but safety is a critical focus for us. We see humanoids as a platform, much like how personal computers revolutionized tasks beyond single-purpose machines. Just as personal computers made tasks more accessible, humanoids can do a variety of tasks in a compact, efficient form.
We draw inspiration from safety standards used in autonomous systems, such as self-driving cars. In the same way that a self-driving car must be safe at high speeds, our humanoid robots must be reliable and safe, especially in environments where failure could cause harm. For example, if a sensor fails or a computer dies, the robot should still operate safely with degraded performance rather than causing damage. We are continually working on improving the robustness and safety of our robots.
Can you share how you met your co-founder and the early days of building UCR?
Wenda: Our CEO, Wei Ding, and I worked together at Apple for about seven years, so we already had a strong working relationship. Wei also had a close friend Wen-loong who was doing humanoid research at Caltech before humanoids became mainstream. This connection led Wei to meet Christopher McQuin, who worked on the humanoid project Valkyrie, a humanoid robot developed by NASA over a decade ago. Between Wei, Wen-loong, and Chris, we had a lot of experience in humanoid research, and that’s how we came together as co-founders.
What were the first few months like for UCR after founding the company?
Wenda: The early days were chaotic, but exciting. We had a vision for humanoid robots, but figuring out our technical approach and how to communicate that vision to investors took time. We joined an incubation program at SPC, which helped us get the ball rolling. In just four months, we were able to build our first humanoid prototype from scratch and even showcased it at a public demo. At SPC’s demo day, people were amazed at how quickly we had developed a working humanoid robot. It was a challenging and intense period, but ultimately very rewarding.
What was the outcome of your demo day, and how did you handle skepticism from investors and competitors in the humanoid space?
Wenda: The demo was a turning point. SPC’s investment and mentorship were key to our progress, and the public demo showed our potential. However, there was skepticism, especially from investors, given the competitive landscape, especially since companies like Boston Dynamics and Agility Robotics are well-established players. But we knew our approach was different. By focusing on heavy industries first, we can build a solid foundation in a more controlled environment, which will later enable us to scale. The challenges were significant, but they helped us refine our focus and prove the value of our approach.
Looking ahead, where do you see UCR in the next year? Any exciting plans?
Wenda: In the next year, we’re focusing on delivering high-quality, reliable robots for heavy industries. Our immediate goal is to get our humanoids to the point where they can perform tasks with exceptional reliability and safety. We’re also looking to expand our team and explore potential collaborations with other companies in the robotics ecosystem. As we continue to improve our technology, we’ll be ready to scale.