Register now After registration you will be able to apply for this opportunity online.
This opportunity is not published. No applications will be accepted.
Model Predicitve Control of a Soft Robotic Actuator System
Semester Project / Master Thesis: Model predictive controller design, hardware deployment and experimental validation for motor position and actuator volume control in a soft robotic actuator system.
Keywords: Model Predictive Control, State Estimation, Soft Robotics, Embedded Systems
Your task in this project will be to develop a model predictive controller (MPC) for an existing soft robotic actuator system that is used in biomedical applications such as the manipulation of physiological pressure waveforms. A sketch of the system is depicted below on the right-hand side. The system’s voice-coil motor exerts a linear force on a bellows that is attached to a water-filled reservoir. The pressure increase in this compartment eventually leads to an outflow of water into the connected end actuator. The volume of the end actuator in turn determines the degree to which the physiological pressure waveform is manipulated.
Controller design for the presented soft robotic actuator system is a challenging but interesting task because it is a single-input multiple-output system with multiple state constraints and non-linear balloon actuator dynamics. The currently deployed cascaded PID controller for motor position control is to be upgraded to a MPC for improved motor position control or even direct end actuator volume control. Rigorous system analysis, MPC design, hardware deployment and testing are the focus of this project, however, the scope of the project can potentially be extended to the investigation of safe learning-based MPC approaches.
**Work packages:**
- MPC design for motor position and actuator volume control
- Implementation and testing of the controller in simulation
- Deployment of the controller on to the real-time computing system
- Experimental evaluation and tuning of the controller
- Optional: Investigation of safe learning-based MPC approaches
Your task in this project will be to develop a model predictive controller (MPC) for an existing soft robotic actuator system that is used in biomedical applications such as the manipulation of physiological pressure waveforms. A sketch of the system is depicted below on the right-hand side. The system’s voice-coil motor exerts a linear force on a bellows that is attached to a water-filled reservoir. The pressure increase in this compartment eventually leads to an outflow of water into the connected end actuator. The volume of the end actuator in turn determines the degree to which the physiological pressure waveform is manipulated.
Controller design for the presented soft robotic actuator system is a challenging but interesting task because it is a single-input multiple-output system with multiple state constraints and non-linear balloon actuator dynamics. The currently deployed cascaded PID controller for motor position control is to be upgraded to a MPC for improved motor position control or even direct end actuator volume control. Rigorous system analysis, MPC design, hardware deployment and testing are the focus of this project, however, the scope of the project can potentially be extended to the investigation of safe learning-based MPC approaches.
**Work packages:** - MPC design for motor position and actuator volume control - Implementation and testing of the controller in simulation - Deployment of the controller on to the real-time computing system - Experimental evaluation and tuning of the controller - Optional: Investigation of safe learning-based MPC approaches