Register now After registration you will be able to apply for this opportunity online.
This opportunity is not published. No applications will be accepted.
FEM-based modeling and control of a sensorized soft pneumatic actuator
Modeling the dynamics of a sensorized fiber-reinforced soft continuum pneumatic arm using a finite-element-based simulation framework. Use the model to control the soft arm in a contact-rich environment.
Keywords: Soft Robotics, Finite Element Method, Robotic Control, Hybrid Control Systems, Object Manipulation, Real-time Control
Soft robots, due to their inherent compliance, are expected to be used in scenarios where traditional, rigid-linked robots cannot be used, such as in close proximity to humans. However, various modelling methods exist to describe the behaviour of the arm, and the most effective modelling method for a soft robot arm remains an open question.
This project aims to develop a FEM-based model of an existing soft pneumatic arm. This soft proprioceptive continuum arm (SoPrA) contains internally flex sensors from Bend Labs, which can measure the curvature state of the robot. An analytical dynamical model that describes the behavior of the robot already exists, but additionally a finite element method simulation-based modeling of the robot may yield an alternative model of the soft arm which can complement the existing analytical model.
In this project, you will e modeling a fiber-reinforced pneumatic actuator in an FEM simulation framework (SOFA or DiffPD). Currently neither of these frameworks supports fiber reinforcement, and thus in its current state it cannot simulate SoPrA. You will have to explore various methods to simulate a fiber-reinforced actuator, considering various simplifications to make the model tractable, in order to make a simulated model that can predict the actual robot’s motion accurately. By having an accurate and efficient simulated model of SoPrA in simulation, it will become possible to apply various FEM-based sensing and control methods. Your next step will be to use this model in proprioceptive sensing. For example, through the discrepancy between the model and measured bending state, the contact state of the arm can be known. Further, you can adapt the methods used in previous research to obtain a reduced-order linear model of the FEM model, and apply linear control methods on the soft arm.
Soft robots, due to their inherent compliance, are expected to be used in scenarios where traditional, rigid-linked robots cannot be used, such as in close proximity to humans. However, various modelling methods exist to describe the behaviour of the arm, and the most effective modelling method for a soft robot arm remains an open question.
This project aims to develop a FEM-based model of an existing soft pneumatic arm. This soft proprioceptive continuum arm (SoPrA) contains internally flex sensors from Bend Labs, which can measure the curvature state of the robot. An analytical dynamical model that describes the behavior of the robot already exists, but additionally a finite element method simulation-based modeling of the robot may yield an alternative model of the soft arm which can complement the existing analytical model.
In this project, you will e modeling a fiber-reinforced pneumatic actuator in an FEM simulation framework (SOFA or DiffPD). Currently neither of these frameworks supports fiber reinforcement, and thus in its current state it cannot simulate SoPrA. You will have to explore various methods to simulate a fiber-reinforced actuator, considering various simplifications to make the model tractable, in order to make a simulated model that can predict the actual robot’s motion accurately. By having an accurate and efficient simulated model of SoPrA in simulation, it will become possible to apply various FEM-based sensing and control methods. Your next step will be to use this model in proprioceptive sensing. For example, through the discrepancy between the model and measured bending state, the contact state of the arm can be known. Further, you can adapt the methods used in previous research to obtain a reduced-order linear model of the FEM model, and apply linear control methods on the soft arm.
- Review relevant papers regarding different soft robot modelling methods
- understand the robot system of SoPrA.
- Model a fiber-reinforced pneumatic actuator (or its approximation) in an FEM-based simulation framework.
- Evaluate the accuracy of the simulated model in predicting actual robot movements
- Combine the FEM model and the proprioceptive sensor of SoPrA to estimate contact states
- Use the simulated model in a reduced-order model-based controller for SoPrA
- Compare different modelling approaches for soft continuum arm robots and discuss advantages and disadvantages for each
- Review relevant papers regarding different soft robot modelling methods - understand the robot system of SoPrA. - Model a fiber-reinforced pneumatic actuator (or its approximation) in an FEM-based simulation framework. - Evaluate the accuracy of the simulated model in predicting actual robot movements - Combine the FEM model and the proprioceptive sensor of SoPrA to estimate contact states - Use the simulated model in a reduced-order model-based controller for SoPrA - Compare different modelling approaches for soft continuum arm robots and discuss advantages and disadvantages for each
- Understanding of basics of finite element method calculations
- Ability to code in Python or C++ (code for SoPrA is mainly C++, but a Python interface is available)
- Understanding of basic mathematical optimization methods
- Understanding of robot kinematics / dynamics
- Cooperation with students / researchers at SRL. There are people with backgrounds in materials science, mechanical / electrical engineering, computer science, physics who can support you on your project.
- Understanding of basics of finite element method calculations - Ability to code in Python or C++ (code for SoPrA is mainly C++, but a Python interface is available) - Understanding of basic mathematical optimization methods - Understanding of robot kinematics / dynamics - Cooperation with students / researchers at SRL. There are people with backgrounds in materials science, mechanical / electrical engineering, computer science, physics who can support you on your project.
Application via SiROP portal with the following documents:
1. Cover letter: your relevant experience in this area, tentative timeline, motivation
2. CV: professional summary, education with GPAs, research and work experience, key skills, publications (if any), awards and fellowships (if any), volunteering work (if any), and outside interests - exemplary publication or thesis work (if available)
3. 2 reference contacts
4. Transcripts of all prior and ongoing degrees
Further information about the Soft Robotics Lab can be found on srl.ethz.ch
Yasunori Toshimitsu, yasu313nori@gmail.com, Robotics Researcher
Prof. Robert Katzschmann, rkk@ethz.ch, Institute of Robotics and Intelligent Systems, D-MAVT
Application via SiROP portal with the following documents:
1. Cover letter: your relevant experience in this area, tentative timeline, motivation 2. CV: professional summary, education with GPAs, research and work experience, key skills, publications (if any), awards and fellowships (if any), volunteering work (if any), and outside interests - exemplary publication or thesis work (if available) 3. 2 reference contacts 4. Transcripts of all prior and ongoing degrees
Further information about the Soft Robotics Lab can be found on srl.ethz.ch