A provider of artificial intelligence (AI)-driven robotics has developed a humanoid robot designed for the industrial sector.
Richtech Robotics has unveiled Dex, the company’s first mobile humanoid robot for industrial use.
Source: Richtech Robotics
Accelerated by the Nvidia Jetson Thor, Dex is capable of operating in dynamic environments, adapting with real-time reasoning and performing complex tasks with detailed precision, all while operating for a full workday on a single charge.
Richtech is using Nvidia technology to accelerate Dex’s training across diverse industrial and commercial contexts.
By combining real-world data with Nvidia Isaac Sim, an open, reference robotics simulation framework, Dex is able to learn tasks virtually at an exponential rate, and implement behaviors into a live industrial environment.
This “Sim2Real” pipeline reduces deployment cycles, enhances safety and enables faster scaling of new robotic applications.
Dex builds on insights from more than 450 Richtech robot deployments nationwide. It combines the autonomous mobile robot (AMR) technology of Richtech’s Titan delivery series with the two-armed precision of the ADAM service robot line, creating a platform that’s both mobile and dexterous.
Richtech made a conscious decision, driven by both engineering and operational efficiency, to opt for wheels rather than legs. Developing a stable, bipedal gait offers no practical benefit in industrial environments.
Legged robots also require substantial power to maintain balance and movement, limiting their runtime to as little as 15 minutes and necessitating multiple units or frequent recharging to remain operational.
In contrast, Richtech’s wheeled AMR platform can carry heavier loads, run for hours on a single charge, and deliver faster braking, tighter maneuvering and greater stability in shared human environments — all while maintaining lower energy and maintenance costs.
Matt Casella, president of Richtech, said: “Humans are great at object manipulation, and wheels are best for fast, efficient, and stable transportation.
“So we designed Dex to travel like a machine and perform tasks like a person. This practical approach gives Dex unmatched versatility and uptime, enabling us to solve real-world production challenges today.”
Dex boasts an eight-hour battery life in mobile mode and can run continuously 24/7 from a static base. Its dual production arms feature modular end-effectors for hands, clamps or specialized tools. A four-camera vision system allows Dex to navigate and perform tasks in a rapidly changing environment.
After Dex completes initial training in simulation, Richtech refines its performance with real-world data. This combination of simulated learning and real-world reinforcement helps Dex adapt faster and perform more reliably in complex environments.
As more companies bring manufacturing back to the U.S., Dex offers a practical path to higher productivity without increasing labor strain. Dex can be trained quickly with new data and adapted for specialized workflows, helping manufacturers scale operations efficiently and sustain domestic production.
To collect real-world data at scale, Richtech launches American robotics data initiative.
Casella said: “For years, America’s robotics industry has been constrained by a lack of regionally grounded data.
“By investing in domestic training and large-scale data collection, we’re empowering the next generation of physical AI to be developed, refined, and led from right here in the U.S.”
The company plans to eventually license portions of this dataset, supporting broader physical AI companies in the U.S.
Dex can handle light or medium industrial tasks, including:
- Machine operation for manufacturing
- Part sorting and material handling
- Quality inspection and packaging
These core capabilities make Dex an immediate productivity partner in manufacturing, logistics and other labor-intensive sectors.
