Advancements in Data-Driven Shape Control for Robotics

This article serves as the second part of a tutorial focused on data-driven, Jacobian-based shape control in robotics. It delves into the deformation Jacobian, which models how robot movements affect the shape of deformable objects.

Key Findings

The authors utilize singular value decomposition to categorize shape changes based on robot motions. They discuss the implications of selecting shape features on the controller's effectiveness in regulating object deformation.

Challenges in Shape Control

The paper identifies three potential failures in the shape control loop: loss of controllability, loss of observability, and mechanical instabilities. Additionally, it presents extensions for managing time-varying targets and three-dimensional objects.