Physical artificial intelligence company MobilTech is building a test and evaluation platform with the Agency for Defense Development (ADD) to verify the performance of military unmanned autonomous vehicles in virtual environments.
MobilTech said on the 12th that it is developing the virtual test platform as part of ADD's project to advance virtual testing environments for the evaluation of unmanned autonomous vehicles. The project runs through April 2028 and aims to cut the costs and safety constraints of physical vehicle testing while improving the reliability of autonomous driving algorithms.
The company will create a digital twin of a roughly 10-kilometer test track at ADD's Changwon Test Center. It will reproduce outdoor driving conditions by reflecting actual road surface undulations and the soil characteristics of unpaved terrain, and will simulate the operation of vehicle sensors such as cameras and LiDAR in the virtual environment. The company also plans to allow testing of autonomous driving performance by setting various weather conditions, including rain, snow and fog, and by linking actual vehicle controllers and software to the virtual environment.
MobilTech will also take part in the Defense AX Hub opening ceremony held the same day in Pangyo, Seongnam, Gyeonggi Province, where it will showcase Replica City, its own digital twin-based virtual test platform. Replica City is a technology that reproduces real terrain and facilities in high-precision three-dimensional environments. The company plans to present simulation projects it has carried out in areas including defense, ports, parking and urban environments, along with cases of physical AI development and verification.
"As military unmanned systems become more advanced, there are constraints such as cost and safety in sufficiently verifying a wide range of terrains and environments through physical testing alone," MobilTech CEO Kim Jae-seung said. "Through digital twin and simulation technology, we will build an environment where autonomous driving functions and performance can be verified before physical testing, and we will advance the test and evaluation infrastructure that supports the development of defense physical AI."