Skip to Content

See and Follow: Object Following using Soft Computing Control Techniques

 

 A control strategy is developed based on visual information given by an adaptive tracking method based on color information. A visual fuzzy servoing system has been developed control a quadcopter (MUAV), that also considers its own dynamics. This system is focused on continuously following of an aerial moving target object, maintaining it with a fixed safe distance and centered on the image plane. The control behavior is validated on real flights on outdoors scenarios, showing the robustness of the proposed systems against winds perturbations that affect not just to the quadcopter but also to the object to follow. 

 

Using cameras in outdoor environments is a challenging task. Sudden changes and inconsistencies with outdoor illumination cause changes in the apparent colour as perceived by a camera. We approach the problem of tracking by exploiting the colour characteristic of the target. We define a basic colour to the target by assuming a simple coloured mark to it and tracking this mark. Therefore, we rely on a suitable and consistent colour representation that allows us to keep colour distributions derived from video image sequences approximately constant (in outdoor settings). However, this process is not always perfect, and changes still occur in colour distributions over time. An algorithm that has proven to deal with this issue by dynamically adapting to changes in probability distributions is the Continuously Adaptive Mean Shift (CamShift).

The Fuzzy controllers were implemented using the Miguel Olivares’ Fuzzy Software(MOFS). For this work two fuzzy controllers were defined, one for control the yaw or heading and the other to acts over the pitch state of the MUAV. Each one has two inputs and one output, being all the variables defined using triangular membership functions. 49 if-then rules are defined for the yaw controller and 30 for the pitch controller.

 

Real outdoors tests have been made with excellent results. The fast response of the controllers permit to follow the moving object (red balloon) suring more than couple of minutes. Some of theses resuls are shown in the figures at the right side of this text.

 

 

 

GPS trajectory reconstruction of the MUAV during a object following test.

GPS trajectory reconstruction of the MUAV during a object following test.

Angle between the MUAV, the object to follow and the centre of the image (input of the yaw controller).

 

Angle between the MUAV, the object to follow and the centre of the image (input of the yaw controller).

Size estimation of the red balloon during the flight (Pitch controller input).

Size estimation of the red balloon during the flight (Pitch controller input).