This module is an example module for the course AE4317 Autonomous Flight of Micro Air Vehicles at the TU Delft.

This module is used in combination with a color filter (cv_detect_color_object) and the guided mode of the autopilot. The avoidance strategy is to simply count the total number of orange pixels. When above a certain percentage threshold, (given by color_count_frac) we assume that there is an obstacle and we turn.

The color filter settings are set using the cv_detect_color_object. This module can run multiple filters simultaneously so you have to define which filter to use with the ORANGE_AVOIDER_VISUAL_DETECTION_ID setting. This module differs from the simpler orange_avoider.xml in that this is flown in guided mode. This flight mode is less dependent on a global positioning estimate as with the navigation mode. This module can be used with a simple speed estimate rather than a global position.

Here we also need to use our onboard sensors to stay inside of the cyberzoo and not collide with the nets. For this we employ a simple color detector, similar to the orange poles but for green to detect the floor. When the total amount of green drops below a given threshold (given by floor_count_frac) we assume we are near the edge of the zoo and turn around. The color detection is done by the cv_detect_color_object module, use the FLOOR_VISUAL_DETECTION_ID setting to define which filter to use.

Example for airframe file

Add to your firmware section: This example contains all possible configuration options, not all of them are mandatory!

<module name="orange_avoider_guided">
  b'<define name="ORANGE_AVOIDER_VISUAL_DETECTION_ID" value="ABI_BROADCAST" />\n    '
  b'<define name="FLOOR_VISUAL_DETECTION_ID" value="ABI_BROADCAST" />\n  '
</module>

Module configuration options

Define Options

name: ORANGE_AVOIDER_VISUAL_DETECTION_ID value: ABI_BROADCAST
Description: which VISUAL_DETECTION message to recieve for orange pole detection.
name: FLOOR_VISUAL_DETECTION_ID value: ABI_BROADCAST
Description: which VISUAL_DETECTION message to recieve for floor detection.

Module functions

Init Functions

These initialization functions are called once on startup.

orange_avoider_guided_init()

Periodic Functions

These functions are called periodically at the specified frequency from the module periodic loop.

orange_avoider_guided_periodic()
- Frequency in Hz: 4
- Autorun: LOCK
  Periodic function automatically starts after init and can't be stopped.

Files

Header Files

The following headers are automatically included in modules.h

orange_avoider_guided.h

Source Files

modules/orange_avoider/orange_avoider_guided.c

Raw orange_avoider_guided.xml file:

<!DOCTYPE module SYSTEM "module.dtd">
 
<module name="orange_avoider_guided" dir="orange_avoider">
  <doc>
    <description>
This module is an example module for the course AE4317 Autonomous Flight of Micro Air Vehicles at the TU Delft.
This module is used in combination with a color filter (cv_detect_color_object) and the guided mode of the autopilot.
The avoidance strategy is to simply count the total number of orange pixels. When above a certain percentage threshold,
(given by color_count_frac) we assume that there is an obstacle and we turn.
 
The color filter settings are set using the cv_detect_color_object. This module can run multiple filters simultaneously
so you have to define which filter to use with the ORANGE_AVOIDER_VISUAL_DETECTION_ID setting.
This module differs from the simpler orange_avoider.xml in that this is flown in guided mode. This flight mode is
less dependent on a global positioning estimate as with the navigation mode. This module can be used with a simple
speed estimate rather than a global position.
 
Here we also need to use our onboard sensors to stay inside of the cyberzoo and not collide with the nets. For this
we employ a simple color detector, similar to the orange poles but for green to detect the floor. When the total amount
of green drops below a given threshold (given by floor_count_frac) we assume we are near the edge of the zoo and turn
around. The color detection is done by the cv_detect_color_object module, use the FLOOR_VISUAL_DETECTION_ID setting to 
define which filter to use.
    </description>
    <define name="ORANGE_AVOIDER_VISUAL_DETECTION_ID" value="ABI_BROADCAST" description="which VISUAL_DETECTION message to recieve for orange pole detection."/>
    <define name="FLOOR_VISUAL_DETECTION_ID" value="ABI_BROADCAST" description="which VISUAL_DETECTION message to recieve for floor detection."/>
  </doc>
  <settings>
    <dl_settings>
      <dl_settings name="OrangeAvoiderGuided">
         <dl_setting var="oag_color_count_frac" min="0" step="0.01" max="0.5"/>
         <dl_setting var="oag_floor_count_frac" min="0" step="0.01" max="0.5"/>
         <dl_setting var="oag_max_speed" min="0" step="0.1" max="1"/>
         <dl_setting var="oag_heading_rate" min="0" step="5" max="45" unit="rad" alt_unit="deg"/>
      </dl_settings>
    </dl_settings>
  </settings>
  
  <dep>
    <depends>cv_detect_color_object</depends>
  </dep>
  <header>
    <file name="orange_avoider_guided.h"/>
  </header>
  <init fun="orange_avoider_guided_init()"/>
  <periodic fun="orange_avoider_guided_periodic()" freq="4"/>
  <makefile target="ap|nps">
    <file name="orange_avoider_guided.c"/>
  </makefile>
</module>