Register now After registration you will be able to apply for this opportunity online.
This opportunity is not published. No applications will be accepted.
Rotor-Failure Recovery MPC
Develop an MPC for quadrotors under rotor failures.
Keywords: MPC mpc model predictive control quadrotor quad rotor failure
A multitude of advanced control techniques for quadrotors have matured over the recent years, with the most promising one being Model Predictive Control. While drones nowadays display high robustness and impressive flight capabilities, they’re not resilient to all possible failures. Especially in the case where a motor failure occurs most controllers struggle to keep the drone stable. Since the quadrotor is already an underactuated vehicle, the loss of one rotor also implies the loss of control over one degree-of-freedom, significantly changing the system dynamics. However, MPC provides some elegant methods to catch such failures.
This thesis will investigate the possibilities in controlling a vehicle under rotor failure, design a control system within an existing flight software stack and use our latest drone hardware.
The requirements are a basic knowledge of control systems, preliminary experience in optimal control, such as MPC, and being familiar with C++ programming.
A multitude of advanced control techniques for quadrotors have matured over the recent years, with the most promising one being Model Predictive Control. While drones nowadays display high robustness and impressive flight capabilities, they’re not resilient to all possible failures. Especially in the case where a motor failure occurs most controllers struggle to keep the drone stable. Since the quadrotor is already an underactuated vehicle, the loss of one rotor also implies the loss of control over one degree-of-freedom, significantly changing the system dynamics. However, MPC provides some elegant methods to catch such failures. This thesis will investigate the possibilities in controlling a vehicle under rotor failure, design a control system within an existing flight software stack and use our latest drone hardware. The requirements are a basic knowledge of control systems, preliminary experience in optimal control, such as MPC, and being familiar with C++ programming.
The goal is to develop a solution using Model Predictive Control to catch a rotor failure, demonstrate it on an existing real quadrotor platform, and compare it to other existing approaches.
The goal is to develop a solution using Model Predictive Control to catch a rotor failure, demonstrate it on an existing real quadrotor platform, and compare it to other existing approaches.
Philipp Föhn (foehn at ifi.uzh.ch),
Sihao Sun (sun at ifi.uzh.ch)
Philipp Föhn (foehn at ifi.uzh.ch), Sihao Sun (sun at ifi.uzh.ch)