Korean Researchers Innovate With Memory-Mimicking Robots To Enhance Industrial Efficiency

Korean researchers have introduced a "Physical AI" technology that enhances autonomous mobile robots' navigation in logistics centres and smart factories. This innovation, developed by Korea's Daegu Gyeongbuk Institute of Science and Technology (DGIST), mimics human forgetting to help robots differentiate between crucial, real-time obstacles and outdated information. The study, published in the Journal of Industrial Information Integration, highlights how this method improves robot efficiency.

Professor Kyung-Joon Park, the study's author, explained that the technology replicates the social principle of forgetting unnecessary data while retaining essential information. This approach allows robots to move more efficiently by focusing only on relevant obstacles. "We have mimicked the social principle of forgetting unnecessary information while retaining only important information to enable efficient movement," said Park.

Autonomous Mobile Robots (AMRs) are prevalent in logistics and manufacturing but often face delays due to temporary obstructions. Traditional systems reroute around these blockages even after they are cleared, reducing productivity. To address this issue, Park’s team implemented a collective intelligence algorithm inspired by human social behaviour.

The new model enables robots to share only vital details like sudden obstructions while naturally forgetting outdated ones. This technique optimises cooperative navigation skills, potentially reducing operating costs, power consumption, and equipment maintenance for companies. The researchers tested their model using the Gazebo simulator in a logistics centre setting.

The results demonstrated significant performance improvements over conventional ROS 2 navigation systems. Average driving time decreased by up to 30.1 percent, and task throughput increased by up to 18 percent. These enhancements suggest substantial savings in energy use and equipment wear.

Broader Applications Beyond Factories

The method requires only 2D LiDAR sensors and is available as a plugin for ROS 2, facilitating easy adoption. Researchers believe this approach could extend beyond factory settings to applications such as logistics robots, autonomous vehicles, drone swarms, smart city traffic management, and large-scale exploration and rescue operations.

By integrating this human-like forgetting mechanism into robotic systems, the technology offers a promising solution for improving efficiency in various sectors. The potential cost savings and enhanced performance make it an attractive option for industries seeking to optimise their operations with advanced robotics solutions.

With inputs from WAM