[E-Daily Reporter Shin Yeong-bin ] SamsungElectronics(005930)has partnered with the University of Michigan to conduct research on physical artificial intelligence (AI) aimed at improving robots’ spatial understanding and interaction with humans.

According to the information technology (IT) industry on the 29th, SamsungElectronics’ U.S.-based advanced R&D subsidiary, Samsung Research America (SRA), recently selected eight final research projects for “START 2026,” an industry-academia collaboration program targeting North American universities.

START is a program operated by SRA to conduct joint research on future technologies with researchers at North American universities. It connects universities’ basic and cutting-edge research with SamsungElectronics’ research and development (R&D) to identify technologies that can be applied to future products and services. Selected projects will receive up to $150,000 in funding over one year. This year, more than 400 research proposals were submitted from approximately 130 universities across North America.

Particularly notable fields this year are robotics and physical AI. In this area, Professor Yulun Tian of the University of Michigan was selected as a finalist. Professor Tian is researching spatial intelligence—which robots need to understand and navigate their surroundings—as well as robot perception, navigation, and multi-robot systems.

SamsungElectronics has also outlined specific research directions for Physical AI. As a core research topic, the company proposed technology that integrates the understanding and memory of 3D space with the prediction of future states into a single AI system. This technology is intended to enable robots to continuously assess their surroundings and predict upcoming situations.

Foundation models for robots that interact closely with humans were also proposed as a research focus. This technology enables robots to go beyond simply executing pre-programmed commands to understanding human behavior and the surrounding environment and moving accordingly.

In the field of agentic AI, a large number of leading North American AI researchers were selected. Among those named were Aviral Kumar, a professor at Carnegie Mellon University; Diyi Yang, a professor at Stanford University; Felix Heide, a professor at Princeton University; and Ranjay Krishna, a professor at the University of Washington.

Professor Kumar is a researcher who has worked in the fields of reinforcement learning and Agent AI. Professor Yang conducts research in natural language processing and human-centered AI. Professor Heide is recognized as an expert in computer vision and computational imaging.

In addition, Professor Jason Wu of Purdue University was selected in the multimedia field. Professor Jianping Pan of the University of Victoria was named in the open category, which seeks to identify innovative research without specific topic restrictions.

In its research proposal, the SRA stated, “We are focused on advancing the cognitive abilities of future robots, such as humanoids and service robots,” adding, “We believe we can accelerate the development of robotics technology through human-like cognitive systems that integrally provide spatial understanding, memory, and future prediction capabilities.”