Addverb Expands Industrial Robotics Workflow Using NVIDIA AI Platforms
As part of the initiative, Addverb is using NVIDIA Omniverse libraries and NVIDIA Cosmos World Foundation Models to create high-fidelity digital twins of warehouses and industrial environments.
Sangeet Kumar, Co-founder & CEO, Addverb
You're reading Entrepreneur India, an international franchise of Entrepreneur Media.
Addverb, a global robotics and warehouse automation firm, has expanded its robotics development workflow by integrating artificial intelligence, simulation, and edge computing technologies from NVIDIA.
The enhanced system is being deployed across the company’s Trakr quadruped robot and the Elixis-W wheeled humanoid platform to accelerate design, training, testing, and large-scale deployment of industrial robots.
As part of the initiative, Addverb is using NVIDIA Omniverse libraries and NVIDIA Cosmos World Foundation Models to create high-fidelity digital twins of warehouses and industrial environments. These physically accurate virtual replicas allow engineers to test robot behavior in simulated real-world conditions using synthetic data. The approach helps identify design and operational issues earlier in development, reducing deployment time and improving reliability.
To improve the transfer of learning from simulation to real-world environments, the company is adopting NVIDIA Isaac Lab, an open-source robot learning framework built on NVIDIA Isaac Sim. The GPU-accelerated platform enables scalable training and evaluation of robot policies and supports reinforcement learning workflows within a unified development stack.
Addverb is also evaluating simulation workflows enabled by Newton, an open-source physics engine co-developed by NVIDIA, Google DeepMind, and Disney Research and managed by the Linux Foundation. In parallel, the company is exploring server-side learning and edge deployment of Vision Language Action (VLA) models using NVIDIA Jetson Thor.
For real-world deployment, Addverb robots are powered by NVIDIA Jetson Orin NX modules, with NVIDIA TensorRT enabling low-latency inference at the edge. This setup supports real-time perception, navigation, and decision-making in complex industrial settings.
The expanded workflow reflects Addverb’s focus on building production-grade Physical AI systems aimed at improving automation efficiency and scalability in warehouses and manufacturing facilities.
Founded in 2016, Addverb develops automation solutions including autonomous mobile robots, sorting systems, storage and retrieval systems, and humanoid robots designed for industrial tasks such as material handling, inspection, and repetitive operations.
Addverb, a global robotics and warehouse automation firm, has expanded its robotics development workflow by integrating artificial intelligence, simulation, and edge computing technologies from NVIDIA.
The enhanced system is being deployed across the company’s Trakr quadruped robot and the Elixis-W wheeled humanoid platform to accelerate design, training, testing, and large-scale deployment of industrial robots.
As part of the initiative, Addverb is using NVIDIA Omniverse libraries and NVIDIA Cosmos World Foundation Models to create high-fidelity digital twins of warehouses and industrial environments. These physically accurate virtual replicas allow engineers to test robot behavior in simulated real-world conditions using synthetic data. The approach helps identify design and operational issues earlier in development, reducing deployment time and improving reliability.
