Uncertainty-Aware Shared Control for Vision-Based Micromanipulation
Tian et al., 2025
Embryologists must precisely control microscopic tools to manipulate individual cells such as sperm and eggs, a task requiring skill and experience.
This study presents a new human–robot collaboration system for micromanipulation, in which a robotic assistant works with, rather than replaces, a human operator. A key feature is that the system can estimate and signal its own uncertainty, essentially “knowing when it’s unsure” and ask for operator assistance.
The approach combines computer vision to track both tools and targets with a shared control framework that lets the operator provide guidance when visual tracking becomes unreliable. In controlled laboratory tests, the method reduced calibration errors, cut the average time to reach a target by more than half, and improved tracking accuracy compared to established techniques.
While not yet tested in real clinical procedures, this development shows promise for more reliable and efficient micromanipulation tools in IVF automation.