Capturing Motion,
Crafting Stories

Teaching robots to catch like humans!

A IROS 2025 research proposed a framework combining optimization and learning for compliant catching on mobile manipulators. It also introduces a P-LSTM network to effectively learn catching strategies from human demonstrations.

NOKOV Motion Capture provided high-accuracy pose data for human demonstrations and algorithm validation.

Results show 98.7% success in simulation and 92.6% in real-world experiments, with a 28.7% reduction in impact torque.

Learn more: https://en.nokov.com/motion-capture-applications/robotics-engineering.html