Korean technology mimics human forgetfulness to enhance robot intelligence.

 

Korean technology

Korean technology mimics human forgetfulness to enhance robot intelligence.

 A team of researchers in the Republic of Korea has developed a new technology that mimics human forgetting to improve the ability of autonomous mobile robots to adapt to complex environments.

The technology, which Daegu Gyeongbuk College of Science and Technology has dubbed "material artificial intelligence," leverages the "spreading and forgetting" principle of human behavior and aims to enable robots to distinguish between truly important obstacles and unnecessary or outdated information by mimicking how human memory works.

Autonomous mobile robots rely heavily on their ability to adjust their path when encountering obstacles such as lift trucks or irregularly shaped cargo. However, in traditional systems, the robot retains old information, causing it to repeat unnecessary paths even after the obstacle has been removed. This reduces productivity, especially in sensitive environments such as logistics centers and smart factories.

To address this problem, the team led by Professor Kyung-Joon Park, the study's author, implemented a collective intelligence algorithm inspired by human social phenomena, mathematically simulating how information spreads rapidly and is gradually forgotten.

This model allows robots to share only important information, such as the location of a sudden obstacle, while ignoring unnecessary or outdated details. This enhances their ability to interact efficiently with the environment, optimize routes, and reduce delays, thereby reducing operating costs, energy consumption, and equipment maintenance.

The team tested the model using the Gazebo simulator to simulate a logistics center environment, and the results showed a significant improvement over the traditional ROS 2 routing system.

The study recorded a 30.1% reduction in average movement time and an 18.0% increase in performance efficiency.

The technology is easy to implement and highly compatible with other systems, requiring only a 2D LiDAR device without the need for additional sensors, which reduces equipment costs. It is also available as a ready-made software package within the ROS 2 system, enabling direct integration into autonomous navigation systems without any hardware changes.

This feature enables the technology to be quickly applied in diverse industrial environments, including logistics robots, self-driving cars, and drone swarms. It is expected to play an important role in broader applications, such as traffic management in smart cities and large-scale search and rescue operations.