In collaboration with NVIDIA and ADI, presenting a digital twin environment for physical AI training and verification

With the growing demand for cable management automation driven by the proliferation of AI data centers, Synopsys has unveiled robot simulation technology for data centers in collaboration with NVIDIA and ADI.

Synopsys, together with Analog Devices and NVIDIA, showcased physical AI simulation technology for the training and performance verification of industrial robots for data center cable management at NVIDIA GTC 2026, held in San Jose, California, from March 16 to 19.

Recently, hyperscale data centers consist of thousands of servers and kilometers of cables, increasing the need for automation in cable connection, replacement, and organization. While cable work may appear to be simple, repetitive tasks, it is a precision operation that requires robots to perform position alignment, force control, and obstacle handling.

To implement this, Synopsys utilized Ansys' simulation technology to build a physics-based virtual environment. This is a method that provides the conditions necessary for robot learning and verification by reflecting the physical characteristics of cables and connectors, such as elasticity, breaking force, and angle limits, in the NVIDIA Isaac Sim environment.

Physical parameters generated by Ansys Mechanical are packaged as OpenUSD assets and integrated into ADI’s Isaac Sim environment. This reduces the gap between the real and virtual environments and enables robots to learn precision tasks, such as cable connections, under more realistic conditions.

Sensor simulation is also utilized. Ansys AVxcelerate Sensors can be applied to verify the performance of ToF sensor-based 3D recognition systems, and based on this, ADI plans to advance the development of robotics solutions, including ToF sensors and tactile sensors.

ADI plans to summarize the results of this collaboration into a robotics benchmark and share it with the industry. The benchmark is focused on supporting companies in starting the development of AI solutions at NVIDIA Isaac Labs by utilizing their own robot platforms or ADI reference designs.

Simulation-based development has industrial significance in that it can reduce the burden of repetitive physical testing and prototype manufacturing. Particularly in fields where it is difficult to replicate large-scale environments, such as data centers, verification methods utilizing synthetic data and digital twins can be used to accelerate robot development.

At GTC 2026, Synopsys demonstrated a dual-arm robotic arm equipped with force, vision, and contact sensing capabilities, along with a corresponding digital twin. This technology has the potential to expand into industrial robot sectors requiring precision operation, such as manufacturing and logistics, as well as data centers.