Korean Scientists Teach Robots to “Forget” for Faster Navigation

In a breakthrough that could reshape factory and warehouse automation, Korean scientists have created a “Physical AI” system that lets robots mimic human forgetting. This innovation aims to make autonomous mobile robots (AMRs) faster and smarter in navigating complex environments.

Unlike conventional systems that stubbornly reroute around obstacles even after they’re gone, the new technology enables robots to distinguish between crucial, real-time blockages and outdated information. This human-inspired forgetting ensures smoother movement, higher efficiency, and less wasted energy.

Stay informed with the latest news. Follow DXB News Network on WhatsApp Channel

The research, conducted by Korea's Daegu Gyeongbuk Institute of Science and Technology (DGIST) and published in the Journal of Industrial Information Integration, uses a collective intelligence algorithm modeled on social memory. Robots can instantly share key information about sudden obstructions while naturally discarding details that no longer matter.

Tests using the Gazebo simulator, which replicated a logistics center, revealed impressive results. Robots using this system reduced average driving time by up to 30 percent and improved task throughput by 18 percent. The method relies on only 2D LiDAR sensors and is available as a ROS 2 plugin, making it easy to implement in existing facilities.

Beyond warehouses, experts suggest the technology could enhance autonomous vehicle routing, drone swarm coordination, smart city traffic management, and even large-scale exploration or rescue missions. With the potential to save energy, reduce maintenance, and cut operational costs, this human-like approach to robot intelligence marks a significant step forward in automation technology.