Mqtt agent for rpi-cam-web-interface
pip install rpicammqtt==0.9.4
Mqtt agent for RPi-Cam-Web-Interface (https://elinux.org/RPi-Cam-Web-Interface) project.
The agent runs on a raspberry pi where RPi-Cam-Web-Interface is installed and running.
The main objective is to provide an mqtt interface to the FIFO file (/var/www/FIFO11) to inject commands and get the status from status_mjpeg.txt.
Additionally, the support for pantilt is also available (check below).
The agent has been developed with the specific purpose to simplify the integration with homeassistant; this is made possible through the dedicated HA camera platform "rpicam" (https://gitlab.com/gbus/ha-rpicam), currently under development. Anyway, the generic nature of the mqtt messages makes it possible to use any mqtt controller to read and send info with the agent.
When connected to an MQTT server it is possible to:
The topics are:
rpicam/<camera_name>/status
, status from /dev/shm/mjpeg/status_mjpeg.txtrpicam/<camera_name>
, topic used to publish commandsrpicam/<camera_name>/pt
, topic used to publish pan/tilt commandsrpicam/<camera_name>/pt/views
, list of configured viewsrpicam/<camera_name>/events/motion
, motion eventsEnsure that the command pip3 is installed and run:
sudo pip3 install rpicammqtt
Install the requirements, clone the repo and run the setup:
sudo apt-get install git python3-setuptools python3-yaml
git clone git@gitlab.com:gbus/rpi-cam-mqtt.git
cd rpi-cam-mqtt
sudo python3 setup.py install
The configuration is looked up in:
/etc/rpi-cam-mqtt/rpi-cam-mqtt.yaml
$HOME/.rpi-cam-mqtt/rpi-cam-mqtt.yaml
Create the preferred directory:
sudo mkdir /etc/rpi-cam-mqtt
or
mkdir $HOME/.rpi-cam-mqtt
cd into it and add the files:
sudo wget https://gitlab.com/gbus/rpi-cam-mqtt/raw/master/rpicammqtt/config/rpi-cam-mqtt.yaml
sudo wget https://gitlab.com/gbus/rpi-cam-mqtt/raw/master/rpicammqtt/config/Adafruit_PCA9685.yaml
In rpi-cam-mqtt.yaml change at least the following:
- Camera name
- MQTT server address
- MQTT user/pw
Configure the service to start at boot:
sudo wget -o /etc/systemd/system/rpi-cam-mqtt.service https://gitlab.com/gbus/rpi-cam-mqtt/raw/master/rpicammqtt/systemd/rpi-cam-mqtt.service
sudo systemctl daemon-reload
sudo systemctl enable rpi-cam-mqtt
sudo systemctl start rpi-cam-mqtt
When RPi-Cam-Web-Interface is set in monitoring mode, a macro is executed every time motion starts or stops.
rpicam
is able to publish these events as MQTT message on a dedicated topic:
rpicam/<camera_name>/events/motion
The macro relies on the netcat command:
sudo apt-get install netcat
Add to the macro path the following script:
cat <webpath>/macros/motion_event.sh
(usually webpath = /var/www/html)
#!/bin/bash
printf $1 | /bin/netcat 127.0.0.1 8888
ensure the script motion_event.sh
is executable.
Pan/Tilt can be enabled and it is based on the PCA9685 controller and the Adafruit library. It is possible to predefine pan/tilt pair values with memorable names so that the camera can be instructed to point to a specific location (ie: point to "front_window"). The configuration looks like this:
pantilt:
views:
front_window:
- 560
- 355
door:
- 420
- 355
The pan/tilt values can be easily identified running the command "configpt", available from your path after the installation of the module.
This allows you to move the arrow keys while watching the live feed from the camera and get the pan/tilt values:
# configpt
Pan: 540 Tilt: 470
When the views are noted down with the correct pan/tilt values press q to exit the command.
An easy way to test the agent is to check the topics for your camera with an mqtt client, ie with mosquitto:
mosquitto_sub -h MQTTSRV_IP -p 1883 -u USERNAME -P PASSWORD -q 1 -t rpicam/CAMERA_NAME/status
mosquitto_sub -h MQTTSRV_IP -p 1883 -u USERNAME -P PASSWORD -q 1 -t rpicam/CAMERA_NAME/pt/views
or all of it:
mosquitto_sub -h MQTTSRV_IP -p 1883 -u USERNAME -P PASSWORD -q 1 -F 'Topic: %t\nMsg: %p\n' -t rpicam/CAMERA_NAME/#
It is possible to send any commands you would normally send to the FIFO file as described in the documentation (https://elinux.org/RPi-Cam-Web-Interface#Pipe), ie:
Disable the camera:
mosquitto_pub -h MQTTSRV_IP -p 1883 -u USERNAME -P PASSWORD -q 1 -t rpicam/CAMERA_NAME -m "ru 0"
Enable motion detection:
mosquitto_pub -h MQTTSRV_IP -p 1883 -u USERNAME -P PASSWORD -q 1 -t rpicam/CAMERA_NAME -m "md 1"
Set a new annotation and add a background to the text:
mosquitto_pub -h MQTTSRV_IP -p 1883 -u USERNAME -P PASSWORD -q 1 -t rpicam/CAMERA_NAME -m "an Entrance hall %Y.%M.%D_%h:%m:%s"
mosquitto_pub -h MQTTSRV_IP -p 1883 -u USERNAME -P PASSWORD -q 1 -t rpicam/CAMERA_NAME -m "ab 1"
When the agent receives the string commands, they are parsed before being sent to the pipe. The filtering is based on the type schema and the limits defined in data/rpi-cam-info.json
The camera can be pointed to different corners by setting values for pan and tilt like this:
mosquitto_pub -h MQTTSRV_IP -p 1883 -u USERNAME -P PASSWORD -q 1 -t rpicam/CAMERA_NAME/pt -m "{\"pan\": 380, \"tilt\": 300}"
Alternatively, one of views can be retreived:
mosquitto_sub -h MQTTSRV_IP -p 1883 -u USERNAME -P PASSWORD -q 1 -t rpicam/CAMERA_NAME/pt/views
...and used:
mosquitto_sub -h MQTTSRV_IP -p 1883 -u USERNAME -P PASSWORD -q 1 -t rpicam/CAMERA_NAME/pt -m "{\"view\": \"front_window\"}"
After the agent is started by systemctl check for process errors:
systemctl status rpi-cam-mqtt
If the process is correctly run, check the logs:
tail -f /var/log/rpi-cam-mqtt.log
Increase the logging level in the configuration if needed:
logging:
level: ERROR|INFO|DEBUG