How Marple is used in drone design

Data analysis has become a major part of research and development in aerospace systems. Aeroplanes, drones, machines, satellites, ... gather more and more data every day. They are stuffed with hundreds, if not thousands of sensors. This data is used to analyse the behaviour of the system and to optmise the performance. In this use case we will look at the development of a drone position control system and how Marple is used to visualise and understand the data.

In this example we will run a simulation of a quadcopter using MATLAB Simulink. The simulation environment allows us to quickly iterate and tune our control system. We can save the data from the simulation in either a .mat or .csv format.

Marple can load data from various sources

A simulink model of a drone

We will run a simulation in which the drone will move to another location and see how the position controller performs. The data is visualised using our specialised time series data visualisation tool.

Marple autoscales signals with different value ranges in the same view

A visualisation with auto-scaled data

We can see that the velocity target is clearly too agressive as the drone does not manage to achieve the target in time. This causes the drone to overshoot the position target as well. Let's run another simulation where we tune down the position controller and see what the effect is.

Overshoot in a controller

Clearly the velocity target is lower now and the drone is able to achieve it. Now we reach the target position without any overshoot. However, the controller is too slow now so let's see if we can find a better balance.

Marple can switch between different files or simulations

The corrected controller

The position control of the quadcopter seems to work perfect now. There is no overshoot while maintaining high performance.

In this example we saw how Marple can be used to analyse and develop a control system. In this case the position control of a quadcopter was tuned in an iterative manner.


With Marple we were able to

  • Easily load the data
  • Plot different signals from the data
  • Quickly use the visualisation on new data
  • Clearly see the relation between different signals