Tesi etd-03212025-125151
Link copiato negli appunti
Tipo di tesi
Dottorato
Autore
DU, CHENGJIN
URN
etd-03212025-125151
Titolo
Soft Tactile-driven Grasping for Intelligent Fruit Handling and Evaluation
Settore scientifico disciplinare
ING-INF/06
Corso di studi
Istituto di Biorobotica - PHD IN BIOROBOTICA
Commissione
Presidente Prof. CUTKOSKY, MARK
Membro Prof. CIANCHETTI, MATTEO
Membro Dott. VISENTIN, FRANCESCO
Membro Prof. CIANCHETTI, MATTEO
Membro Dott. VISENTIN, FRANCESCO
Parole chiave
- Tactile Sensing
- Soft Robotics
- Sensorized Gripper Systems
- Fruit Handling and Evaluation
Data inizio appello
17/07/2025;
Disponibilità
parziale
Riassunto analitico
This thesis presents the development of a tactile-enhanced soft robotic gripper designed for delicate fruit handling. The work explores how integrating magnetic tactile sensing with soft actuators can enable adaptive, gentle, and informed grasping.
The first part of the thesis introduces the background and motivation, highlighting the emergence of soft robotics as a promising alternative to rigid manipulation in tasks requiring compliance. It further identifies the challenges of tactile feedback in soft systems, particularly in agricultural contexts where fruits of varying firmness and ripeness must be handled without damage.
A key technical contribution is the design and development of the SoftMag sensor, a soft magnetic tactile sensor based on a Hall-effect sensing mechanism. A theoretical model incorporating magnetostatics and elastomer deformation is developed to inform the sensor design, followed by numerical optimization and simulation. The fabricated sensor is then extensively characterized under single- and multi-touch conditions, showing sensitivity to both normal and shear forces with a favorable signal-to-noise ratio.
The sensor is then integrated into soft pneumatic actuators to create sensorized fingers, enabling localized tactile feedback. Through a series of tests—including tactile calibration, actuation, kinematics, and dynamic performance evaluation—the integrated system demonstrates decent real-time performance. A payload handling test is also conducted, demonstrating a maximum grasping capacity of over 800 grams.
Building upon this, a complete soft sensorized gripper is developed with embedded tactile feedback (implemented by a neural network) and compact mechatronic architecture. In particular, the mechanical parasitic effect of the system is explicitly identified, addressed, and evaluated. Followed by the demonstration of slippage detection based on closed-loop control methodology. As a specific application in the fruit industry, a firmness estimation system is developed within the system framework and evaluated by extensive probing experiments. Further, a redesign focuses on miniaturization and integration with a robotic arm, validating the gripper’s practical applicability in real-world scenarios.
The thesis concludes with a critical discussion of the system’s capabilities, comparing it to existing soft grippers and highlighting improvements in sensitivity, integration, and tactile-enhanced grasping. Limitations, such as localized sensing and environmental robustness, are acknowledged. Future directions include distributed sensing, electromagnetic compatibility, advanced control strategies, and deployment in agricultural environments.
Additional explorations are presented in the appendices, with an elephant-inspired soft gripper and a prototype “super sensor”—a visual-magnetic dual modal tactile sensor, demonstrating the extensibility of the core ideas of bridging soft robotics with tactile intelligence.
The first part of the thesis introduces the background and motivation, highlighting the emergence of soft robotics as a promising alternative to rigid manipulation in tasks requiring compliance. It further identifies the challenges of tactile feedback in soft systems, particularly in agricultural contexts where fruits of varying firmness and ripeness must be handled without damage.
A key technical contribution is the design and development of the SoftMag sensor, a soft magnetic tactile sensor based on a Hall-effect sensing mechanism. A theoretical model incorporating magnetostatics and elastomer deformation is developed to inform the sensor design, followed by numerical optimization and simulation. The fabricated sensor is then extensively characterized under single- and multi-touch conditions, showing sensitivity to both normal and shear forces with a favorable signal-to-noise ratio.
The sensor is then integrated into soft pneumatic actuators to create sensorized fingers, enabling localized tactile feedback. Through a series of tests—including tactile calibration, actuation, kinematics, and dynamic performance evaluation—the integrated system demonstrates decent real-time performance. A payload handling test is also conducted, demonstrating a maximum grasping capacity of over 800 grams.
Building upon this, a complete soft sensorized gripper is developed with embedded tactile feedback (implemented by a neural network) and compact mechatronic architecture. In particular, the mechanical parasitic effect of the system is explicitly identified, addressed, and evaluated. Followed by the demonstration of slippage detection based on closed-loop control methodology. As a specific application in the fruit industry, a firmness estimation system is developed within the system framework and evaluated by extensive probing experiments. Further, a redesign focuses on miniaturization and integration with a robotic arm, validating the gripper’s practical applicability in real-world scenarios.
The thesis concludes with a critical discussion of the system’s capabilities, comparing it to existing soft grippers and highlighting improvements in sensitivity, integration, and tactile-enhanced grasping. Limitations, such as localized sensing and environmental robustness, are acknowledged. Future directions include distributed sensing, electromagnetic compatibility, advanced control strategies, and deployment in agricultural environments.
Additional explorations are presented in the appendices, with an elephant-inspired soft gripper and a prototype “super sensor”—a visual-magnetic dual modal tactile sensor, demonstrating the extensibility of the core ideas of bridging soft robotics with tactile intelligence.
File
| Nome file | Dimensione |
|---|---|
Ci sono 1 file riservati su richiesta dell'autore. |
|