Mobilint is moving into development of on-device AI semiconductors for robots and home appliances. Mobilint said on Sept. 28 it will participate in four projects under the Ministry of Trade, Industry and Energy's "K-On-Device AI Semiconductor Technology Development Programme" and cooperate with LG Electronics, Doosan Robotics and Daedong. The programme is a national R&D project in which AI chip companies and demand-side companies jointly develop and validate semiconductors for products by industry.

Mobilint will lead detailed AI chip development tasks for projects involving LG Electronics' humanoid, Doosan Robotics' collaborative robot and Daedong's unmanned farming robot. It will join the AI home project with LG Electronics as a joint R&D organisation. Under the model, demand-side companies present the requirements needed for their products, and Mobilint reflects them in chip development and validates the chips on devices. Unlike the previous approach of completing a chip first and then looking for customers, the company will align requirements from the early stages of development.

For the LG Electronics humanoid project, it will develop an on-device AI accelerator to help process AI computations needed to recognise and judge the surrounding environment inside the device. For the Doosan Robotics collaborative robot project, it will jointly develop an AI semiconductor for next-generation collaborative robots and an SDK, a software development tool, and confirm applicability on an actual robot platform. For the Daedong project, it will be responsible for an AI semiconductor for unmanned farming robots and building a development environment.

Mobilint plans to develop computation processing technology tailored to each robot's functions and operating conditions. The company said it has supported customers in running AI models on actual devices using its own NPU and software development environment. In this programme, it will develop semiconductors and software environments according to the scope of each project and validate them in connection with the demand-side companies' devices. It plans to reflect validation results in technology advancement and follow-on commercialisation.

Mobilint CEO Dong-ju Shin (신동주) said, "The competitiveness of a domestic NPU depends not only on chip performance but on proving performance and economic feasibility in actual products and services." He said, "Through these four projects, we will pursue development of AI semiconductors tailored to demand-side companies' product requirements, as well as validation on real devices and commercialisation."