Follow the guide given below, but first: Note that the version of ardupilot you have to clone into your system has changed. Use this: https://github.com/sdhar04/Agri_Loiter-Stable guide Make sure you are installing Ubuntu 20.04. Dual Booting FAQs:
- Can I run things on a VM instead? MacBooks cannot be dual booted, and VMs generally run better on them. In case of Windows systems, it is strongly advised to dual boot as VMs are significantly slower and computationally inefficient.
- The device shows an error after grub, saying boot device not found. Reinstall Ubuntu, this time change the location of boot device to some other drive, but take absolute care so it doen't interfere with Windows.
In a new terminal launch the runway.world using the code below.
roslaunch iq_sim runway.launch
startsitl.sh is a shell file that contains the commands to start the SITL simulation, with arguments that depend on the iris drone provided by Gazebo models. The SITL in itself will connect automatically to the iris drone which is included in the world file.In a new terminal run the following:
./startsitl.sh
This results in a screen like this:
In a seperate terminal launch the following command to launch mavros, mavros helps the companion board communicate with the flight control generally in GUIDED mode but in our custom mode, since we are using some part of GUIDED mode we need this.
roslaunch mavros apm.launch fcu_url:=udp://:14550@
In the startsitl terminal, enter mode guided. Press enter a few times so that the terminal reflects the change. Now do:
arm throttle
takeoff 10
Once the drone has reached a stable height, open a new terminal and run realnew.py
python3 realnew.py
Observe the drone starts to takeoff, remember the above takeoff command is only applicable for our custom mode or GUIDED mode.
This algorithm has been tested on 2D LiDAR data in simulation. The next goal is to make it work with 3D data, which may be coming from LiDAR or RaDAR sensors. The main idea is to take heights on distances of obstacles in front of and below the drone. We have to do a few angle corrections as the drone will not always stay perfectly level. Once we get arrays from both LiDARs, we concatenate them and find heights and distances of all detections. The heights are multiplied with a weight that depends on the drone's current forward velocity. Higher velocity means farther obstacles are given higher priority. Initially a 1/v dependence was used, but it was later changed to ln(k.v). Out of these weighted heights, the most notable one is taken to publish a target height, with an offset of 3m.
The equation used was
- self.k1math.log(self.k2self.f_vel))*math.pow(x-d-3,3))+1)*h_values[i]
In the PID loop used, Kp has been made dependent on velocity and distance:
- self.Kp = 0.6*math.exp(self.f_vel)/self.fard
Video of its application:
alt_maintain8.mp4
A potential field function has also been used to limit horizontal velocity in case of very tall obstacle, so that the drone gets enough time to complete its ascent.
The custom mode can be invoked inside SITL using mode 29.
In this mode, te pilot has zero control over the z-axis of the drone, thrust is computed by the companion board. When no input is given, the drone keeps its current position.
To run that, refer to this repo