# Real-World Autonomy: At Production Scale | Rasmus Bendvold | TEDxDetroit Source: https://www.youtube.com/watch?v=lyg6hzozwV0 Recap page: https://rapidrecap.app/video/lyg6hzozwV0 Generated: 2026-02-02T16:33:44.198+00:00 --- ## Quick Overview Rasmus Bendvold, Managing Director US for wheel.me, argues that the future of automation should focus on retrofitting existing equipment rather than building entirely new, specialized robotics, contrasting the spectacular but specialized capabilities of humanoid robots with the practical utility of autonomous drive systems that augment current industrial assets to solve real-world labor gaps. **Key Points:** - The presentation addresses the manufacturing labor gap, projected to grow from 500,000 open positions today to 2 million by 2030, according to Deloitte. - Bendvold contrasts spectacular, specialized robots (like the Boston Dynamics Atlas robot doing a backflip) with the practical utility of autonomous systems designed for real-world tasks. - He advocates for a 'Retrofit, Don't Rebuild' approach, suggesting that augmenting existing carts and equipment with autonomous capabilities is more efficient and cost-effective than designing new, complex robots. - The core concept of Physical AI is defined as Artificial Intelligence embedded in robots, machines, and devices that sense, think, and act in the physical world, following a 'See, Decide, Do' loop. - The wheel.me technology allows existing industrial carts (like one weighing 700 pounds) to become autonomous by replacing their wheels, enabling them to perform tasks like delivering materials in high-traffic areas. - The benefit of this augmentation is improved working conditions, greater efficiency, and a high return on investment for customers. ![Screenshot at 00:24: Rasmus Bendvold is introduced as the Managing Director US for wheel.me, setting the stage for his discussion on practical, scaled autonomy solutions.](https://ss.rapidrecap.app/screens/lyg6hzozwV0/00-00-24.jpg) **Context:** Rasmus Bendvold, Managing Director US for wheel.me, delivers a TEDxDetroit talk titled "Real-World Autonomy: At Production Scale." He sets the context by highlighting a significant projected labor shortage in the manufacturing sector in the US, citing statistics from Deloitte, and argues for a practical approach to automation that integrates with existing infrastructure rather than relying solely on building complex, specialized machines. ## Detailed Analysis Rasmus Bendvold argues that the focus for automation should shift from creating spectacular but specialized robots to implementing practical, scalable solutions that leverage existing industrial assets—a philosophy summarized as "Retrofit, Don't Rebuild" (07:55). He immediately addresses the looming manufacturing labor shortage, noting that 500,000 positions are currently open, expected to rise to 2 million by 2030 (03:59). Bendvold contrasts the impressive but perhaps impractical feats of humanoid robots, like Atlas doing a backflip (00:49), with the utility of Physical AI, which he defines as AI embedded in physical devices that Sense, Decide, and Do (05:15). He asserts that the industry needs to focus on utility over spectacle (01:11). The solution presented by wheel.me involves retrofitting existing equipment, such as a 700-pound cart, by replacing its standard wheels with autonomous drive units (08:12). This allows the existing cart to autonomously navigate warehouses, pick up materials, move them across areas with forklift traffic, and deliver them on demand (07:04). The key advantage is that this approach does not require replacing the entire system; only the wheels are swapped, making the process less complex, less costly, and instantly improving working conditions and efficiency for employees, leading to a high return on investment. ### The Labor Problem - 500k open manufacturing positions today, projected to reach 2 Million by 2030 (as estimated by Deloitte in 2024) - This growing gap demands solutions that scale quickly and effectively. ### Physical AI Defined - The core concept is Artificial Intelligence embedded in robots, machines, and devices that Sense, Decide, and Do (05:15) - This loop emphasizes practical action over mere spectacle. ### Spectacle vs. Utility - Humanoid robots performing backflips are impressive but specialized (00:49) - Real-world problems require utility, such as moving heavy, awkward loads safely and reliably. ### The Retrofit Solution - Bendvold champions 'Retrofit, Don't Rebuild' (07:55) - wheel.me replaces the wheels on existing industrial carts (like a 700lb cart) with autonomous drive units. ### Industrial Application - The retrofitted cart autonomously navigates the warehouse, picks up materials, navigates forklift traffic, and delivers on demand (07:04) - This improves working conditions, efficiency, and provides a huge ROI. ### Conclusion - The future involves creating better jobs that enable people to be more creative and solve complex problems, rather than simply automating away all manual tasks (09:40). ![Screenshot at 00:05: The presentation begins with dynamic, abstract visuals of colorful particles exploding, setting an energetic tone.](https://ss.rapidrecap.app/screens/lyg6hzozwV0/00-00-05.jpg) ![Screenshot at 00:21: The speaker, Rasmus Bendvold, stands on stage in front of a large screen displaying a split image of a Boston Dynamics Atlas robot performing a backflip in two different environments.](https://ss.rapidrecap.app/screens/lyg6hzozwV0/00-00-21.jpg) ![Screenshot at 00:37: The title slide appears: "Real-World Autonomy: At Production Scale" over a dark teal background.](https://ss.rapidrecap.app/screens/lyg6hzozwV0/00-00-37.jpg) ![Screenshot at 01:52: A slide contrasts the complexity of humanoid robots \(left\) with the utility of autonomous systems, showing the Atlas robot doing a backflip.](https://ss.rapidrecap.app/screens/lyg6hzozwV0/00-01-52.jpg) ![Screenshot at 03:59: A slide quantifying the labor gap: 500k open manufacturing positions today vs. an estimated 2 Million by 2030, according to Deloitte.](https://ss.rapidrecap.app/screens/lyg6hzozwV0/00-03-59.jpg)