Precise In-Hand Manipulation of Soft Objects with Soft Fingertips | RoboSoft 2020

Added:

Rigid Manipulators with High-Resolution Tactile Sensing

While soft pneumatic fingertips offer natural compliance, they suffer from material hysteresis, slow response times, and highly non-linear dynamics that complicate precise mathematical modeling. A prominent counter-perspective advocates for rigid robotic manipulators equipped with high-resolution tactile sensors (such as vision-based GelSight sensors). Proponents of this rigid-body approach argue that precise in-hand manipulation is best achieved through exact kinematic control, high control-loop bandwidth, and dense tactile feedback. Rigid systems provide the predictability and repeatability necessary for micro-positioning tasks. Unlike soft fingertips, which can deform unpredictably under varying payloads and are susceptible to puncture or wear, rigid systems with advanced control algorithms can dynamically estimate contact forces and object geometry. This allows them to manipulate delicate or soft objects with high precision without sacrificing the speed, durability, and analytical tractability inherent to rigid-body mechanics.

Foundations of Soft Robotics: Understanding the mechanics of compliant materials (such as elastomers) and how they differ from traditional rigid-bodied robots.
Principles of Robotic Grasping and In-Hand Manipulation: Familiarity with contact mechanics, friction, and kinematic constraints during object manipulation.
Pneumatic Actuation and Deformation: Basic concepts of how fluidic or air pressure is utilized to actuate and deform soft elastomer structures.
Tactile Sensing in Robotics: Understanding how sensors measure pressure, strain, and deformation to provide feedback during contact.
Closed-Loop Control in Soft Robotics: Studying advanced control algorithms that leverage real-time pressure and tactile feedback for dynamic grasp adjustments.
Machine Learning for Soft Manipulation: Exploring reinforcement learning and data-driven models to manage the high degrees of freedom inherent in soft actuators.
Advanced Fabrication of Soft Sensors: Investigating multi-material 3D printing and casting techniques to seamlessly embed air cavities and miniature sensors into soft materials.
Industrial and Medical Applications: Applying soft manipulation techniques to delicate tasks such as agricultural harvesting, surgical robotics, and fragile object logistics.
474 views11likes11:57@ImperialREDSLabOriginal Release: 2020-05-18

This video presents a novel dual-purpose soft fingertip design that combines tactile sensing and active shape-changing capabilities for precise in-hand manipulation of soft objects. The design uses embedded inflatable air cavities within silicone fingertips, where internal air pressure changes correlate with external forces (via the ideal-gas law), enabling both tactile sensing and controlled deformation. The system achieves translation and rotation of grasped objects by selectively inflating different air cavities—both left cavities for rightward translation, both right cavities for offset translation, and coordinated inflation for rotation. Characterization tests establish mathematical relationships between pressure, deformation, and manipulation performance. Experiments demonstrate that initial object rotation angles affect manipulation range, with optimal performance achieved when air cavities maintain contact with objects during inflation. The closed-loop pressure feedback control ensures constant counter-forces between fingertips and objects, preventing compression while enabling precise manipulation. This approach significantly enhances the dexterity of soft robotic grippers for handling delicate soft objects.