Identification and Active Exploration of Deformable Object Boundary Constraints through Robotic Manipulation

Area of Science:

• Robotics
• Mechanical Engineering
• Computer Science

Background:

• Accurate estimation of deformable object deformation is crucial for robotic manipulation, especially in medical applications.
• Current methods require knowledge of object constitutive parameters and boundary constraints for precise tissue response prediction.

Purpose of the Study:

• To present a novel algorithm for estimating boundary constraints of deformable objects using robotic manipulation data.
• To enhance the identification algorithm with an active exploration technique for improved data acquisition.

Main Methods:

• Utilized tissue deformation data captured by a vision system.
• Employed a multi-stage hill climbing procedure to estimate object boundary constraints.
• Incorporated an information maximization approach for active exploration.

Main Results:

• Successfully estimated boundary constraints for deformable objects from robotic manipulation data.
• Demonstrated the effectiveness of the active exploration technique in extending the identification algorithm.
• Analyzed the impact of uncertainties on the proposed methods through simulations.

Conclusions:

• The developed algorithm provides a novel approach to estimating boundary constraints for deformable objects.
• The findings contribute to more accurate robotic motion planning in applications involving soft tissues.
• The study validates the proposed methods through simulations and experimental evaluations.