Robot Masters String Puppetry
The fictional character Pinocchio starts his fantastic journey as a wooden puppet animated by his creator, the talented Geppetto. If Pinocchio’s story were written today, all the sophisticated string-pulling that so artistically endows him with humanlike movement might just as well be performed by a robot.
A group of researchers from ETH Zurich has introduced a robotic system that can perform animation of real-world string puppets, aka marionettes. “PuppetMaster” operates puppets using the traditional method of control bars attached by wires or strings to a puppet hanging below. The ETH robot can animate a bounding puppy, a slithering snake, or a flying Chinese dragon.
Mastering marionettes could take a human performer years of dedication and repetitive practice. The process poses an even more difficult challenge for robots because first, a typical marionette is driven by different forces, such as gravity, tension forces from human operation via strings, and internal forces; and second, there is no intuitive linkage between the actions of a puppeteer and the motions performed by the marionette.
The PuppetMaster system comprises three main components: a kinematic description of the robot puppeteer; a custom-designed, string-driven marionette that the robot can operate; and a target marionette motion the system is tasked to reproduce.
A key innovation of this paper is researchers’ leveraging of derivatives of motions generated by a forward dynamics simulation to predict how the robot’s actions will affect the motion of the marionette. Researchers employed second-order sensitivity analysis to express the relationship between the motion trajectory of the marionette and of the robot, and built a physics-based simulation model to help the marionette and the robot achieve the targeted motions.
A physical robot — the human-sized, dual arm ABB YuMi® IRB 14000 — was used to evaluate the system. Researchers began with a few preliminary tests such as placing a pendulum inside a cup while avoiding an obstacle. They then gradually raised the level of difficulty to tasks such as animating a flying swallow. Throughout the process, researchers optimized robot design elements such as handle configuration to enable the pair to better realize the target motion.
While this research has laid a great foundation for future robotic puppeteering endeavours, researchers admit there are still areas for improvement, such as differences between simulation and real-world results. The range and subtleties of a marionette’s motions also greatly depend on how the robotic system is designed. So far, the robot can only perform iterative motions in a loop, and researchers want to explore open-loop marionettes in the future.
The implication behind robotic puppeteering extends beyond entertainment. Improving robot dexterity to the level of human hands opens up the potential for automating various types of work such as folding clothes at home or bedding in hotels, wielding a sledge hammer at a construction site, and so on.
Read the paper PuppetMaster: Robotic Animation of Marionettes for more information.
Journalist: Tony Peng | Editor: Michael Sarazen
Need a comprehensive review of the past, present and future of modern AI research development? Trends of AI Technology Development Report is out!
2018 Fortune Global 500 Public Company AI Adaptivity Report is out!
Purchase a Kindle-formatted report on Amazon.
Apply for Insight Partner Program to get a complimentary full PDF report.
Follow us on Twitter @Synced_Global or daily AI news!
We know you don’t want to miss any stories. Subscribe to our popular Synced Global AI Weekly to get weekly AI updates.
