**5. Hardware design**

**Figure 12** depicts the hardware design of the quadrocopter (**Figure 13**). The brain of the system is composed of two processing units, an AVR 32 bit MCU (microcontroller unit) UC3A and the LP‐180 providing an AMD‐x86 processor and 2 × 1.6 GHz system clock [28].

The CPUs can be seen in the centre of the picture. The MCU interfaces all sensors except those connected via USB and performs the control part with real‐time computing, while the task of the LP‐180 contains all functions with a high computational burden such as object recognition and mapping.

The quadrocopter uses a couple of sensors for obstacle detection and is capable of distance‐ controlled collision avoidance [29]. For object recognition, the C270 camera from Logitech is used [30]. All processing is done on‐board the quadrocopter, so it is capable of a fully auton‐ omous flight.

Autonomous Quadrocopter for Search, Count and Localization of Objects http://dx.doi.org/10.5772/63568 15

**Figure 12.** Hardware design.

Furthermore, the dominating colour within the detected circle is calculated by combining several of the most frequent red, green and blue values. It is notable that the average values are not used because they can be heavily influenced by bright spots on the search target which emerge because of unfavourable lighting conditions. The final target colour does not consist of exactly one set of RGB values but of ranges for each colour channel which are derived from

Furthermore, the algorithm searches for more than one radius. Therefore, the initially detected radius ±2 can be chosen as a target range to compensate for light variations of height during the flight. To allow bigger chances in flying altitude, the radius range would have to be adjusted

The actual search is performed using a resolution of 192 × 144 pixels. This allows quick processing while still preserving all the information necessary for a successful detection.

After taking a picture it is converted into a greyscale image and the Hough detection is performed. The number and quality of detected circles heavily depend on the threshold used during the Hough circle detection. A good value for the required number of votes is 30% of the circumference of the smallest radius. With significantly higher values, target objects tend to get missed far too often because the constant movement of the quadrocopter tends to prevent

All detected circles, i.e., all target candidates, are then analysed for their colour. For each pixel inside a candidate's enclosing circle, it is checked if its RGB values lie within a certain range. The range is determined during the initial scan. For a candidate to get confirmed as a target, a certain percentage of its pixels has to be target pixels. Setting this threshold to about 40% was

**Figure 12** depicts the hardware design of the quadrocopter (**Figure 13**). The brain of the system is composed of two processing units, an AVR 32 bit MCU (microcontroller unit) UC3A and

The CPUs can be seen in the centre of the picture. The MCU interfaces all sensors except those connected via USB and performs the control part with real‐time computing, while the task of the LP‐180 contains all functions with a high computational burden such as object recognition

The quadrocopter uses a couple of sensors for obstacle detection and is capable of distance‐ controlled collision avoidance [29]. For object recognition, the C270 camera from Logitech is used [30]. All processing is done on‐board the quadrocopter, so it is capable of a fully auton‐

the LP‐180 providing an AMD‐x86 processor and 2 × 1.6 GHz system clock [28].

according to the currently measured distance of the quadrocopter to the floor.

the original values.

14 Recent Advances in Robotic Systems

*4.3.2. Search*

a good value here.

and mapping.

omous flight.

**5. Hardware design**

the camera from taking sharp pictures.

**Figure 13.** AQopterI8 picture.
