Design of a flight controller to achieve improved fault tolerance

Abstract

In the last years, multirotor aerial vehicles have gained popularity both as consumer products and in professional applications. Safety is one of the main concerns during operation, and different approaches to fault tolerance have been proposed and continue to be developed. For a control system to be able to handle off-nominal situations, failures must be properly detected and identified; therefore, a fault detection and identification algorithm is required. Also, the control loop has to be accordingly modified to cope with each particular failure in the best way possible. These algorithms usually run on the vehicle’s low-level flight computer, imposing on it a large additional computational load. In this work, a fault detection and identification module is used to evaluate its impact in terms of additional processing time on a flight computer based on the Cortex-M3 microcontroller. While a highly optimized version of the algorithm is able to run, it still suggests potential hardware limitations for expanding the system capabilities. The evaluation of the same module on an improved flight computer design based on a Cortex-M7 micro-processor shows a significantly reduced footprint in the overall performance, allowing for the addition of an augmented method for faster failure detection.

Keywords

• flight computer
• unmanned aerial vehicles
• fault tolerance
• fault detection and identification