Register now After registration you will be able to apply for this opportunity online.
This opportunity is not published. No applications will be accepted.
Passivity-Based MPC for Aerial Interaction
Aerial interaction with Micro Aerial Vehicles (MAVs) requires a complex controller that can handle uncertainties that arise during an interaction task (e.g. drilling holes, painting walls). In this project, we aim to combine Passivity-Based and Model Predictive Control to provide robustness while ex
In recent years, passivity based control (PBC) has emerged as a promising control method that provides robustness in uncertain working conditions. Until now, different approaches of passivity have been employed on a few robots, including multi-joint robotic arms [1] and flying robots for aerial interaction [2-4].
PBC follows the idea to design the controller in such a way that the entire closed-loop system is passive, i.e. that it only dissipates energy. As passivity and stability are strongly linked, this control design can give guarantees about its stability (up to a certain degree of model uncertainties).
However, due to the necessity to be passive, PBC can result in overly conservative (cautious) control inputs. Therefore, we aim to combine PBC with Model Predictive Control (MPC). We hope to exploit the predictive quality of MPC to allow the system to momentarily deviate from purely passive control inputs while keeping the system passive in the long term. As a result, we could perform complex interaction tasks that might be too unsafe with standard controllers.
In this project, the student will go through the entire process from designing until evaluating a new controller on a real platform. For the experimental validation we use our custom-built drone which we can equip with end effectors for various interaction tasks.
[1] - Schindlbeck, C., & Haddadin, S. (2015). Unified passivity-based Cartesian force/impedance control for rigid and flexible joint robots via task-energy tanks
[2] Rashad, R., Califano, F., & Stramigioli, S. (2019). Port-Hamiltonian Passivity-Based Control on SE(3) of a Fully Actuated UAV for Aerial Physical Interaction Near-Hovering
[3] Yüksel, B., Secchi, C., Bülthoff, H. H., & Franchi, A. (2019). Aerial physical interaction via IDA-PBC
[4] Acosta, J., de Cos, C. R., & Ollero, A. (2020). Accurate control of Aerial Manipulators outdoors. A reliable and self-coordinated nonlinear approach
In recent years, passivity based control (PBC) has emerged as a promising control method that provides robustness in uncertain working conditions. Until now, different approaches of passivity have been employed on a few robots, including multi-joint robotic arms [1] and flying robots for aerial interaction [2-4]. PBC follows the idea to design the controller in such a way that the entire closed-loop system is passive, i.e. that it only dissipates energy. As passivity and stability are strongly linked, this control design can give guarantees about its stability (up to a certain degree of model uncertainties). However, due to the necessity to be passive, PBC can result in overly conservative (cautious) control inputs. Therefore, we aim to combine PBC with Model Predictive Control (MPC). We hope to exploit the predictive quality of MPC to allow the system to momentarily deviate from purely passive control inputs while keeping the system passive in the long term. As a result, we could perform complex interaction tasks that might be too unsafe with standard controllers. In this project, the student will go through the entire process from designing until evaluating a new controller on a real platform. For the experimental validation we use our custom-built drone which we can equip with end effectors for various interaction tasks.
[1] - Schindlbeck, C., & Haddadin, S. (2015). Unified passivity-based Cartesian force/impedance control for rigid and flexible joint robots via task-energy tanks
[2] Rashad, R., Califano, F., & Stramigioli, S. (2019). Port-Hamiltonian Passivity-Based Control on SE(3) of a Fully Actuated UAV for Aerial Physical Interaction Near-Hovering
[3] Yüksel, B., Secchi, C., Bülthoff, H. H., & Franchi, A. (2019). Aerial physical interaction via IDA-PBC
[4] Acosta, J., de Cos, C. R., & Ollero, A. (2020). Accurate control of Aerial Manipulators outdoors. A reliable and self-coordinated nonlinear approach
- Familiarization with previous work
- Literature study on PBC and MPC
- Theoretical design of the controller
- Implementation in simulation (e.g. Simulink, Gazebo)
- Experimental evaluation on our platform
- Familiarization with previous work - Literature study on PBC and MPC - Theoretical design of the controller - Implementation in simulation (e.g. Simulink, Gazebo) - Experimental evaluation on our platform
- High motivation and interest in the topic
- Methodological and goal-oriented working behavior
- Strong background in control
- Experience programming in C++, knowledge of ROS
- High motivation and interest in the topic - Methodological and goal-oriented working behavior - Strong background in control - Experience programming in C++, knowledge of ROS
Application via SiROP portal with the following documents: - Cover letter - Detailed CV - Transcripts of all prior and ongoing degrees
Further information about the Autonomous Systems Lab under asl.ethz.ch.
Application via SiROP portal with the following documents: - Cover letter - Detailed CV - Transcripts of all prior and ongoing degrees
Further information about the Autonomous Systems Lab under asl.ethz.ch.