This is the main branch for the Distributed Climate Control project
By Dan Coen and Conor Sheridan
Contact: dancoen2@gmail.com / conorksheridan@gmail.com
Usage Instructions for running our code
Our project has only been tested with Windows 8.1 and 10. If using Linux, see config differences in arduinoCommands.py. Additionally, PyKinect2 requires the Windows Kinect SDK v2. There may be more changes necessary to make the code compatible. We are also using python 2.7.13
Should already be installed if on Raspbian
Use Lepton3 branch (or other if using a different sensor)
https://opencv.org/releases.html
or for pip install (may be updated version)
https://pypi.python.org/pypi/opencv-python
sudo apt-get install python-numpy
(if you are using the same pi as us (raspi 3) you shouldn't need to change config settings)
-
Change the raspi configuration settings to enable SPI, VNC, and SSH. Depending on what other sensors you are using you may need to enable I2C.
-
Connect an ethernet cable from the Windows machine to the pi. NO CHANGES should need to be made to network preferences.
-
Use
PING raspberrypion windows machine to ensure the hostname is resolvable on your pc. -
Install VNC viewer on your windows machine.
-
Connect to the pi on VNC viewer using hostname: raspberrypi password: None
You can also use SSH instead of VNC viewer but VNC is useful if you need to make changes.
- On the pi, navigate to where the project has been cloned. install the necessary dependencies and run
python ~/projects/sd-18-slamtown/src/server/thermal_server.py
This should automatically configure a the local ip address of the pis eth0 interface and create the thermal image server on it.
This should be all you need to do on the pi. We found this to be the easiest way to get the networking up and running (no internet sharing/bridge connections, etc.) if you need to commit changes you will have to connect the pi to the internet seperately or mess around with the other methods listed above.
To test if the machine is correctly receiving images, install dependencies and run
python client/thermal_client.py
If you are using the same UP board as us, then everything here should already be present
Download the 32-bit installer here
pip install pykinect2
For PyKinect2
pip install comtypes
For PyKinect2
pip install numpy
For various things
pip install imutils
Used for image reshaping
pip install opencv-python
pip install opencv-contrib-python
This is required to allow usage of keypoint detectors SIFT and SURF which are utilized in panorama stitching, and were removed in OpenCV 3
The Kinect must be plugged into a USB 3.0 port in order to work.
To test to make sure opencv and pykinect are installed, uncomment the bottom of src/kinect/PyKinectFrameGrabber.py and run.
Pyserial Download The com port in src/arduino/arduinoCommands.py must be configured differently for linux.
Attach tilt servo to shield servo pin 1 and pan servo to shield pins 2. When doing so, the black wire should be on the side nearest to the power/usb input.
If you're using a laptop, you can probably get away with not using the shield as well as the external power. The servos typically aren't drawing as much power as they are capible of, so the USB could actually be enough. However, if you are using the UP board, the external power is very necessary, as it does not provide sufficient power.
Then, you should compile and upload the motorcontroller/MotorControl.ino sketch to your arduino (requires the arduino ide). Take note of the com port if you are having issues (should be com4). You can use arduinocommands/arduinoCommands.py to test the motors are working properly by inputting numbers on the keyboard.
Once everything is set up and connected, you should be able to run python integrationtest.py and have everything work. Feel free to send us an email if you have any questions!