Skip to content

awslabs/aws-iot-greengrass-component-kvs-webrtc

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

6 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 

AWS IoT Greengrass V2 Component to send video over Kinesis Video Streams with WebRTC

Using this project, you can create an AWS IoT Greengrass V2 (GGv2) Component that will acquire video from a Video for Linux (v4l) device, open an Amazon Kinesis Video Streams (KVS) with WebRTC signaling channel as 'Master', and when a 'Viewer' connects to that signaling channel, publish the acquired video. This project uses GStreamer to acquire, process, and encode video. The 'pipeline' and source can easily be changed or customized to suit different video sources, processing, or formats.

project architecture

  1. Using this guide, a GGv2 Component, wrapping a Docker container, is created on your local Development host and uploaded to an S3 Bucket
  2. The Component is deployed to the Target Device, running Greengrass Nucleus, which will manage (Install, Start, Stop) the Component
  3. Video from v4l devices -- such as camera attached the Target Device -- is acquired by the Component and published over KVS-WebRTC
  4. Viewers can connect to the KVS-WebRTC channel and view the video stream.

The component is packaged as a Docker container to help ensure portability and minimize conflicts with system libraries, packages, and other configurations. As noted below, take care to match the Instruction Set Architecture (ISA)--e.g. x86 or Arm--of the system that builds and packages the container with the target.

This project targets Linux hosts and was developed using Linux and Mac desktop environments.

This project will guide through

  1. Validation of a v4l video source such as a USB camera or capture card
  2. Build the Container and test locally on a GGv2 installation
  3. Package and publish the GGv2 Component

Prerequisites

Part 1 - Validate Setup, connections, drivers, etc.

This component is built to use a Video for Linux (v4l) device as source. In Part 2, you will build the Docker container with GStreamer and the KVS Application. Device access is provided by passing the specific device to the container in the component recipe. First, validate the device is connected properly and that the ggc_user will have access to the device.

  1. Verify v4l device

On the target Greengrass core, verify the desired v4l device exists

ls -l /dev/video*
#crw-rw----+ 1 root video 81, 0 Mar  9 08:44 /dev/video0
#crw-rw----+ 1 root video 81, 1 Mar  9 08:44 /dev/video1

Your devices may differ in number, etc. Also note the group ownership for your desired device (e.g. /dev/video0)--typically video.

  1. Check the video device format

If your system does not have v4l2-ctl, install it with sudo apt install v4l2-utils.

v4l2-ctl --list-formats-ext --device /dev/video0
# ioctl: VIDIOC_ENUM_FMT
#        Type: Video Capture
#
#        [0]: 'MJPG' (Motion-JPEG, compressed)
#                Size: Discrete 1920x1080
#                        Interval: Discrete 0.017s (60.000 fps)
#        [1]: 'YUYV' (YUYV 4:2:2)
#                Size: Discrete 1920x1080
#                        Interval: Discrete 0.017s (60.000 fps)

NOTE: Your output may be different from the above output. Verify that MJPG format is one of the available outputs. If not, it may be possible to adjust the GStreamer pipeline to perform an appropriate transformation.

  1. Add the ggc_user to the video group.
# change this command if necessary for different group or GGv2 user
sudo usermod -aG video ggc_user
  1. Verify Docker

On the target Greengass core, verify Docker functionality.

sudo docker run hello-world
# verify the phrase "Hello from Docker!" in the output

(Optional) Verify Docker on the build machine, if used.

  1. Add the ggc_user to the docker group.
# modify the user if your GGv2 installation has a different user account
sudo usermod -aG docker ggc_user

Part 2 - Build the Docker Container and test locally

This guide follows a 'local shadow' development method to Create Components where development is done directly on a Greengrass core and then kept in sync with an Amazon Simple Storage Service (S3) bucket for later use and deployment. However, as resources (disk space, RAM, CPU) may be more limited on your target core device, the Docker component can be built on a different system with the same Instruction Set Architecture (ISA).

  1. Edit Dockerfile

Set the TZ variable in the Dockerfile as desired for the location of your Greengrass Core--or the desired localization (e.g. UTC). Available localizations can be found in the /usr/share/zoneinfo/ directory.

ENV TZ=<your timezone, e.g. America/Los_Angeles>
# ENV TZ=America/Los_Angeles

Save the Dockerfile.

  1. Build the image
docker build --rm -t <name> .
# example
#docker build --rm -t kvs .
  1. Test the container

Testing is optional -- BUT RECOMMENDED. If you're happy with the container, you can proceed to package and deploy.

(Optionally), if you built the image on a different system than your Greengrass core, transfer the image to the target machine.

# EXAMPLE - OPTIONAL - transfer image to target
IMAGE_NAME=<image-name>
#IMAGE_NAME=kvs
docker save $IMAGE_NAME > $IMAGE_NAME.tar

# transfer the image to the target... 
# for example, using scp
#scp $IMAGE_NAME.tar target:

# ON THE TARGET
docker load -i <image>.tar

A customized version of the kvsWebRTCClientMasterGstreamerSample was built during the docker build and set as the ENTRYPOINT. This application uses the following environment variables:

variable usage
$AWS_ACCESS_KEY_ID key id for credentials to use KVS
$AWS_SECRET_ACCESS_KEY secret key for key id
$AWS_SESSION_TOKEN (Optional) Session Token if used
$AWS_KVS_CACERT_PATH path to certs -- Dockerfile set this to /certs
$AWS_KVS_LOG_LEVEL 3 or other level from README as desired

For local testing, put these values in a .env file and pass it to the container.

## use values from downloaded credentials
echo "\
AWS_ACCESS_KEY_ID=<your key id>
AWS_SECRET_ACCESS_KEY=<your secret key>
# if using cloud9 or a role, you may also have a session token
#AWS_SESSION_TOKEN=<session token>
AWS_KVS_CACERT_PATH=../certs/cert.pem
AWS_KVS_LOG_LEVEL=3 
" >.env

Pass the environment to the docker container with --env-file, then test the container and the pipeline

CHANNEL_NAME=<kvs-webrtc-channel-name>
#CHANNEL_NAME=test
docker run --env-file ./.env --device=/dev/video0 kvs $CHANNEL_NAME

NB: by default, the KVS application will send VIDEO ONLY, to send both audio and video append audio-video to the above docker command.

Part 3 - Package and Publish as GGv2 Component

It is common practice when building Greengrass components to maintain a 'shadow' of the artifacts and recipes under user home. This guide continues that practice (shadowing components in ~/GreengrassCore in the following commands) as it makes some of the preparation convenient (notably sync'ing file trees to S3). Other workflows are possible.

  1. archive the Docker image
# keeping a local copy of artifacts is generally helpful
mkdir -p ~/GreengrassCore && cd $_

export component_name=<name for your component>
export component_version=<version number>
# example
# export component_name=com.example.kvs_master
# export component_version=1.0.0

# use the name of your docker container created in Part 1
mkdir -p ~/GreengrassCore/artifacts/$component_name/$component_version

export container_name=<name of your container>
# example
# export container_name=kvs

docker save $container_name > ~/GreengrassCore/artifacts/$component_name/$component_version/$container_name.tar
  1. (Optional) remove the original image and reload. This is a good way to verify docker save and load functionality with the container.
docker image ls $container_name
# check the output

docker rmi -f $container_name

# recheck images
docker image ls $container_name
# should be empty set

docker load -i ~/GreengrassCore/artifacts/$component_name/$component_version/$container_name.tar

# and the container should now be in the list
docker image ls

# if desired, clean up intermediate containers
#docker system prune
  1. upload script artifacts to S3
# compress the file first, gzip, xz, and bzip are supporteed by Docker for load
gzip ~/GreengrassCore/artifacts/$component_name/$component_version/$container_name.tar

export bucket_name=<where you want to host your artifacts>
# for example
# export region='us-west-2'
# export acct_num=$(aws sts get-caller-identity --query "Account" --output text)
# export bucket_name=greengrass-component-artifacts-$acct_num-$region

# create the bucket if needed
aws s3 mb s3://$bucket_name

# and copy the artifacts to S3
aws s3 sync ~/GreengrassCore/ s3://$bucket_name/
  1. create the recipe for the component
mkdir -p ~/GreengrassCore/recipes/
touch ~/GreengrassCore/recipes/$component_name-$component_version.json

# paste these values into the editor, next
echo $component_name " " $component_version " " $bucket_name

# edit using IDE or other editor
# for example: vim
# vim ~/GreengrassCore/recipes/$component_name-$component_version.json

And enter the following content for the recipe, replacing paste_bucket_name_here with the name of the bucket you created earlier. Also replace component-name, component-version, and container-name. NOTE you can get the building machine's architecture with uname -m.

If your Greengrass user name is other than ggc_user or if the v4l device is owned by a group other than video, be sure to modify the Install script.

Likewise, if your video device is NOT /dev/video0, change the Startup script to remap your device number to video0.

{
  "RecipeFormatVersion": "2020-01-25",
  "ComponentName": "<component-name>",
  "ComponentVersion": "<component-version>",
  "ComponentDescription": "A component that runs a Docker container to send video from a v4l device to KVS.",
  "ComponentPublisher": "Amazon",
  "ComponentConfiguration": {
      "DefaultConfiguration": {
          "channel": "TestChannel"
      }
  },
  "Manifests": [
    {
      "Platform": {
        "os": "linux",
        "architecture": "<arch_of_machine_building_the_image>"
      },
      "Lifecycle": {
        "Install": {
          "Script": "usermod -aG video ggc_user; docker load -i {artifacts:path}/<container-name>.tar.gz"
        },
        "Startup": {
          "Script": "docker run --device=/dev/video0:/dev/video0 --name=<container-name> kvs {configuration:/channel}"
        },
        "Shutdown": {
          "Script": "docker stop <container-name>"
        }
      },
      "Artifacts": [
        {
          "URI": "s3://<paste_bucket_name_here>/artifacts/<component-name>/<component-version>/<container-name>.tar.gz"
        }
      ]
    }
  ]
}

(Optional) Validate your changes in the JSON file to avoid an errors in creating the component (commonly reported as missing rights to null)

cat ~/GreengrassCore/recipes/$component_name-$component_version.json | jq
# and fix any errors that may be reported
  1. create the GG component with
aws greengrassv2 create-component-version --inline-recipe fileb://~/GreengrassCore/recipes/$component_name-$component_version.json

FINISHED - Next Steps

You have now created a Greengrass V2 Docker component and uploaded it to your AWS Account. Note that when deploying the component, the channel name can be remapped. It is also possible to refactor the recipe to move the video device description to a configurable option.

Testing without a Camera? Install v4l loopback device

If you are using a system without a camera, such as a cloud instance, or as part of a CI/CD pipeline, or perhaps just to use a known, consistent video feed, set up a loopback device.

The v4l2loopback device is used to inject known video content into the /dev/video0 device (using GStreamer). As this device uses a kernel module, it is not containerizable, but should work in a virtual machine. It has been validated to work on an EC2 instance.

Install and configure the Loopback device on the host

These instructions assume an Ubuntu 20.04 or similar host

sudo apt-get update && sudo apt-get upgrade -y

Install V4L2 utils

sudo apt-get install v4l-utils

Build loopback kernel module

cd ~
# or other directory where you keep your source

# clone latest release tag -- v0.12.7 as of this writing
git clone -b v0.12.7 https://github.com/umlaeute/v4l2loopback.git
cd v4l2loopback

make

sudo make install
sudo depmod -a

sudo modprobe v4l2loopback

configure the loopback device

The device must be configured after rebooting or when you wish to change configuration -- when changing resolution, for example.

# load the module after a reboot
sudo modprobe v4l2loopback

v4l2-ctl -d /dev/video0 -c timeout=3000 
v4l2loopback-ctl set-timeout-image /dev/video0 testcard.png  
# ignore ERROR about pipeline not wanting to run
sudo v4l2loopback-ctl set-fps /dev/video0 30
sudo v4l2loopback-ctl set-caps /dev/video0 "UYVY:1280x720@30/1" 

Inject test video with GStreamer

Sample Video

Test feeds are available from the D2 City data set, which is widely cited and can be downloaded from (https://www.scidb.cn/en/detail?dataSetId=804399692560465920).

This section was developed by capturing a hour of YouTube 'dashcam' videos and encoding to MP4 - ~/dashcam.mp4. To use another file, change the reference in the command below as needed.

Run GStreamer Pipeline

in a new window, ensure the v4l loopback device is setup, then

gst-launch-1.0 uridecodebin uri=file:///home/ubuntu/dashcam.mp4 ! \
  decodebin name=demux demux. ! videoconvert ! \
  videoscale ! 'video/x-raw,format=UYVY,width=1920,height=1080,framerate=30/1' ! \
  v4l2sink device=/dev/video0

If your loopback device is other than /dev/video0, modify the above command appropriately.

Troubleshooting

To fix a failed deployment:

Go to Deployments in the console and remove the offending component from the deployment (check both thing and group level). Deploy. This will remove the component from the target.

Delete the component definition in the console

Update the artifacts and push to S3

Re-Create the component definition (as this will take a hash from the artifacts). (alternatively, it should be possible to create a new version)

Add the newly, re-created component to the deployment and deploy.

It can be very handy to turn off the Rollback feature on failure to see what was captured/expanded

If you find yourself iterating through the above cycle many times, it may be easier to develop the component locally first and then upload it. See Create custom AWS IoT Greengrass components for information about how to work with components locally on the Greengrass core.

Verify USB device for capture card

lsusb -t
# /:  Bus 02.Port 1: Dev 1, Class=root_hub, Driver=xhci_hcd/7p, 5000M
#    |__ Port 4: Dev 2, If 0, Class=Video, Driver=uvcvideo, 5000M
#    |__ Port 4: Dev 2, If 1, Class=Video, Driver=uvcvideo, 5000M
#    |__ Port 4: Dev 2, If 2, Class=Audio, Driver=snd-usb-audio, 5000M
#    |__ Port 4: Dev 2, If 3, Class=Audio, Driver=snd-usb-audio, 5000M
# /:  Bus 01.Port 1: Dev 1, Class=root_hub, Driver=xhci_hcd/8p, 480M
#    |__ Port 1: Dev 2, If 1, Class=Human Interface Device, Driver=usbhid, 1.5M
#    |__ Port 1: Dev 2, If 0, Class=Human Interface Device, Driver=usbhid, 1.5M
#    |__ Port 5: Dev 3, If 0, Class=Audio, Driver=snd-usb-audio, 12M
#    |__ Port 5: Dev 3, If 1, Class=Audio, Driver=snd-usb-audio, 12M
#    |__ Port 5: Dev 3, If 2, Class=Audio, Driver=snd-usb-audio, 12M
#    |__ Port 5: Dev 3, If 3, Class=Human Interface Device, Driver=usbhid, 12M
#    |__ Port 6: Dev 4, If 0, Class=Wireless, Driver=btusb, 12M
#    |__ Port 6: Dev 4, If 1, Class=Wireless, Driver=btusb, 12M

NOTE a webcam is found on Bus 02.Port 1, specific bus mapping will vary. Two Video and Two audio interfaces is normal and correct. Your devices may be different.

If using a non-USB device, validate connectivity as appropriate.

Verify V4L2 mapping and function

ls -l /dev/v4l/by-path/
# lrwxrwxrwx 1 root root 12 Aug 12 09:42 pci-0000:00:15.0-usb-0:4:1.0-video-index0 -> ../../video0
# lrwxrwxrwx 1 root root 12 Aug 12 09:42 pci-0000:00:15.0-usb-0:4:1.0-video-index1 -> ../../video1

Note two video capture devices matching the USB devices. If you have additional devices, they may be listed here as well. Do take note of the symbolic link for the device you wish to use (/dev/video0 in this guide). Verify that the path (e.g. pci-0000:00:15.0-usb-0:4:1.0-video-index0 is correct for the physical connection of your device).

Now check the video device binding

ls -l /dev/video*
# crw-rw----+ 1 root video 81, 0 Aug 12 09:42 /dev/video0
# crw-rw----+ 1 root video 81, 1 Aug 12 09:42 /dev/video1

install v4l2 if needed

sudo apt install v4l-utils

check that v4l2 got the device binding

v4l2-ctl --list-devices
Failed to open /dev/video0: Permission denied

and FAIL need to add the user to the video group

sudo usermod -aG video $USER

need to reload the shell to access the new group

and now we see the devices

v4l2-ctl --list-devices
# UVC Camera (1e4e:7016) (usb-0000:00:15.0-4):
#        /dev/video0
#        /dev/video1
#        /dev/media0

Check the device capabilities

v4l2-ctl --device=/dev/video0 --all
# Driver Info:
#        Driver name      : uvcvideo
#        Card type        : UVC Camera (1e4e:7016)
#        Bus info         : usb-0000:00:15.0-4
#        Driver version   : 5.15.39
#        Capabilities     : 0x84a00001
#                Video Capture
#                Metadata Capture
#                Streaming
#               Extended Pix Format
#                Device Capabilities
#        Device Caps      : 0x04200001
#                Video Capture
#                Streaming
#                Extended Pix Format
# Media Driver Info:
#        Driver name      : uvcvideo
#        Model            : UVC Camera (1e4e:7016)
#        Serial           : 20000130041415
#        Bus info         : usb-0000:00:15.0-4
#        Media version    : 5.15.39
#        Hardware revision: 0x00000100 (256)
#        Driver version   : 5.15.39
# Interface Info:
#        ID               : 0x03000002
#        Type             : V4L Video
# Entity Info:
#        ID               : 0x00000001 (1)
#        Name             : UVC Camera (1e4e:7016)
#        Function         : V4L2 I/O
#        Flags         : default
#        Pad 0x01000007   : 0: Sink
#          Link 0x02000010: from remote pad 0x100000a of entity 'Extension 4': Data, Enabled, Immutable
# Priority: 2
# Video input : 0 (Camera 1: ok)
# Format Video Capture:
#        Width/Height      : 1920/1080
#        Pixel Format      : 'YUYV' (YUYV 4:2:2)
#        Field             : None
#        Bytes per Line    : 3840
#        Size Image        : 4147200
#        Colorspace        : sRGB
#        Transfer Function : Rec. 709
#        YCbCr/HSV Encoding: ITU-R 601
#        Quantization      : Default (maps to Limited Range)
#        Flags             : 
# Crop Capability Video Capture:
#        Bounds      : Left 0, Top 0, Width 1920, Height 1080
#        Default     : Left 0, Top 0, Width 1920, Height 1080
#        Pixel Aspect: 1/1
# Selection Video Capture: crop_default, Left 0, Top 0, Width 1920, Height 1080, Flags: 
# Selection Video Capture: crop_bounds, Left 0, Top 0, Width 1920, Height 1080, Flags: 
# Streaming Parameters Video Capture:
#        Capabilities     : timeperframe
#        Frames per second: 30.000 (30/1)
#        Read buffers     : 0

Grab a test frame

v4l2-ctl --device /dev/video0 --set-fmt-video=width=1920,height=1080,pixelformat=MJPG --stream-mmap --stream-to=/tmp/output.jpg --stream-count=1
# <     # this '<' is the normal output

# verify grab -- size and date/time
ls -l /tmp/output.jpg
# -rw-rw-r-- 1 scott scott 41288 Aug 12 10:17 /tmp/output.jpg

Let's look at it!

Convert from MJPG to something viewable...

May need to install ffmpeg

sudo apt install ffmpeg -y

convert

ffmpeg -i /tmp/output.jpg -bsf:v mjpeg2jpeg frame.jpg

open/transfer/view frame.jpg

Mount the target core remotely with sshfs

mkdir target
sshfs <user>@<ip>: target
# unmount with umount

now you can browse that jpg file remotely... Does it look okay?

If evaluating a new camera and need to check format compatibility -- ffmpeg has a mapping of v4l2 modes.

# check output format compatability
ffmpeg -f video4linux2 -list_formats all -i /dev/video0
# ...
# [video4linux2,v4l2 @ 0x557afa242680] Compressed:       mjpeg :          Motion-JPEG : 1920x1080
# [video4linux2,v4l2 @ 0x557afa242680] Raw       :     yuyv422 :           YUYV 4:2:2 : 1920x1080
#

There will likely be a lot of other output, but we are interested in the format lines as shown above. The customized pipelines in this guide use the MJPG format. It may be possible to use YUYV as well as other formats, but those were not tested and would need some modifications to the gStreamer pipelines.

v4l2-ctl --device /dev/video0 --set-fmt-video=width=1920,height=1080,pixelformat=YU12 --stream-mmap --stream-to=/tmp/output.yuv --stream-count=1
# <

Installing the YUV viewer extension to VS Code be helpful here if experimenting with non-MJPG formats.

All-In-One Test

This GStreamer pipeline will fetch audio and video and write an MPEG-4 file. If the MP4 file is accurate, all video and audio devices are configured and functioning properly.

# video only
gst-launch-1.0  -e  \
  v4l2src device=/dev/video0 ! queue ! jpegdec ! \
  videoscale ! video/x-raw,width=1280,height=720 ! \
  videorate ! video/x-raw,framerate=30/1 ! \
  videoconvert ! x264enc bframes=0 speed-preset=veryfast bitrate=512 \
    byte-stream=TRUE tune=zerolatency ! \
  filesink location=test.mp4 sync=false

# with an audio device
gst-launch-1.0  -e  \
  v4l2src device=/dev/video0 ! queue ! jpegdec ! \
  videoscale ! video/x-raw,width=1280,height=720 ! \
  videorate ! video/x-raw,framerate=30/1 ! \
  videoconvert ! x264enc bframes=0 speed-preset=veryfast bitrate=512 \
    byte-stream=TRUE tune=zerolatency ! \
  mux. alsasrc device=hw:1 ! queue ! audioconvert ! audioresample ! \
  voaacenc ! aacparse ! qtmux name=mux ! \
  filesink location=test.mp4 sync=false

NB: If your device does not supply MJPG formatted images, explore available GStreamer plugins or other resources to transform one of the formats that the x264enc plugin will accept (e.g. video/x-raw).

Testing and Troubleshooting Audio

Audio is handled by ALSA. You may need to customize the pipeline above and in the code as described below to reference the correct v4l and ALSA devices.

Available audio devices can be shown with

arecord -l
#**** List of CAPTURE Hardware Devices ****
#card 0: U0x1e4e0x7016 [USB Device 0x1e4e:0x7016], device 0: USB Audio #[USB Audio]
#  Subdevices: 1/1
#  Subdevice #0: subdevice #0
#card 1: KTUSBAUDIO [KT_USB_AUDIO], device 0: USB Audio [USB Audio]
#  Subdevices: 1/1
#  Subdevice #0: subdevice #0

If you get a permissions error in executing arecord, you may need to add the current user to the audio group.

sudo usermod -aG audio $USER

In the above example, note that card 0 is referenced as a USB device with the Manufacturer and Device IDs (0x14e:0x7016). This can be matched with the expected device.

lsusb
#Bus 002 Device 002: ID 1e4e:7016 Cubeternet
#...

For the alsasrc GStreamer plugin, this card is referenced as hw:0. This command will capture raw audio and can be used to test.

arecord -f S16_LE -D hw:0 -c 2 -r 48000 ~/rec0.wav

Note the setting of the format S16_LE along with 2 channels, -c 2, and sample rate, -r 48000. This information can be found in the /proc/asound tree.

cd /proc/asound

cat cards
# 0 [U0x1e4e0x7016  ]: USB-Audio - USB Device 0x1e4e:0x7016
#                      USB Device 0x1e4e:0x7016 at usb-0000:00:15.0-3,super speed
# 1 [KTUSBAUDIO     ]: USB-Audio - KT_USB_AUDIO
#                      KTMicro KT_USB_AUDIO at usb-0000:00:15.0-5, full speed
# 2 [PCH            ]: HDA-Intel - HDA Intel PCH
#                      HDA Intel PCH at 0x91310000 irq 130

cat card0/stream0
#USB Device 0x1e4e:0x7016 at usb-0000:00:15.0-3, super speed : USB Audio
#
#Capture:
#  Status: Stop
#  Interface 3
#    Altset 1
#    Format: S16_LE
#    Channels: 2
#    Endpoint: 0x8a (10 IN) (ASYNC)
#    Rates: 48000
#    Data packet interval: 1000 us
#    Bits: 16
#    Channel map: FL FR

Testing WebRTC

Open the WebRTC Test Page, paste in Access Key Id, Secret Access Key, and CHANNEL_NAME, then Start Viewer.