openpilot-tools

Repo which contains tools to facilitate development and debugging of openpilot.

Table of Contents

• Requirements
• Setup
• Tool examples
  • Replay driving data
  • Debug car controls
  • Stream replayed CAN messages to EON
  • Stream EON video data to a PC
• Welcomed contributions

Requirements

openpilot-tools and the following setup steps are developed and tested on Ubuntu 16.04 and Python 2.7.

Setup

1. Install native dependencies

   • core tools

     sudo apt install git curl python-pip
     sudo pip install --upgrade pip==9.0.3

   • ffmpeg (tested with 3.3.2)

     sudo apt install ffmpeg libavformat-dev libavcodec-dev libavdevice-dev libavutil-dev libswscale-dev libavresample-dev libavfilter-dev

   • build tools

     sudo apt install autoconf automake clang clang-3.8 libtool pkg-config build-essential

   • libarchive-dev (tested with 3.1.2-11ubuntu0.16.04.4)

     sudo apt install libarchive-dev

   • qt python binding (tested with python-qt4, 4.11.4+dfsg-1build4)

     sudo apt install python-qt4

   • capnp 0.6.1 (required for replay)

     curl -O https://capnproto.org/capnproto-c++-0.6.1.tar.gz
     tar xvf capnproto-c++-0.6.1.tar.gz
     cd capnproto-c++-0.6.1
     ./configure --prefix=/usr CPPFLAGS=-DPIC CFLAGS=-fPIC CXXFLAGS=-fPIC LDFLAGS=-fPIC --disable-shared --enable-static
     make -j4
     sudo make install
     
     cd ..
     git clone https://github.com/commaai/c-capnproto.git
     cd c-capnproto
     git submodule update --init --recursive
     autoreconf -f -i -s
     CFLAGS="-fPIC" ./configure --prefix=/usr/local
     make -j4
     sudo make install

   • zmq 4.2.3 (required for replay)

     curl -LO https://github.com/zeromq/libzmq/releases/download/v4.2.3/zeromq-4.2.3.tar.gz
     tar xfz zeromq-4.2.3.tar.gz
     cd zeromq-4.2.3
     ./autogen.sh
     ./configure CPPFLAGS=-DPIC CFLAGS=-fPIC CXXFLAGS=-fPIC LDFLAGS=-fPIC --disable-shared --enable-static
     make
     sudo make install

2. Clone openpilot if you haven't already

   git clone https://github.com/commaai/openpilot.git
   cd openpilot

3. Clone openpilot_tools within openpilot, and install dependencies

   git clone https://github.com/commaai/openpilot-tools.git openpilot_tools
   cd openpilot_tools
   # sudo pip install if not using a venv
   pip install -r requirements.txt
   pip install -r ../requirements_openpilot.txt

4. Add openpilot to your PYTHONPATH.

   For bash users:

   echo 'export PYTHONPATH="$PYTHONPATH:<path-to-openpilot>"' >> ~/.bashrc
   source ~/.bashrc

5. Add some folders to root

   sudo mkdir /data
   sudo mkdir /data/params
   sudo chown $USER /data/params

6. Try out some tools!

Tool examples

Replay driving data

Hardware needed: none

unlogger.py replays data collected with chffrplus or openpilot.

You'll need to download log and camera files into a local directory. Download these from the footer of the comma explorer or SCP from your device.

Usage:

python replay/unlogger.py <route-name> <path-to-data-directory>

#Example:

#python replay/unlogger.py '99c94dc769b5d96e|2018-11-14--13-31-42' /home/batman/unlogger_data

#Within /home/batman/unlogger_data:
#  99c94dc769b5d96e|2018-11-14--13-31-42--0--fcamera.hevc
#  99c94dc769b5d96e|2018-11-14--13-31-42--0--rlog.bz2
#  ...

# In another terminal you can run a debug visualizer:
python replay/ui.py   # Define the environmental variable HORIZONTAL is the ui layout is too tall

Debug car controls

Hardware needed: panda, giraffe, joystick

Use the panda's OBD-II port to connect with your car and a usb cable to connect the panda to your pc. Also, connect a joystick to your pc.

joystickd.py runs a deamon that reads inputs from a joystick and publishes them over zmq. boardd.py sends the CAN messages from your pc to the panda. debug_controls is a mocked version of controlsd.py and uses input from a joystick to send controls to your car.

Usage:

python carcontrols/joystickd.py

# In another terminal:
selfdrive/boardd/boardd.py # Make sure the safety setting is hardcoded to ALL_OUTPUT

# In another terminal:
python carcontrols/debug_controls.py

Stream replayed CAN messages to EON

Hardware needed: 2 x panda, debug board, EON.

It is possible to replay CAN messages as they were recorded and forward them to a EON.  Connect 2 pandas to the debug board. A panda connects to the PC, the other panda connects to the EON.

Usage:

# With MOCK=1 boardd will read logged can messages from a replay and send them to the panda.
MOCK=1 openpilot_tools/replay/boardd.py

# In another terminal:
python replay/unlogger.py <route-name> <path-to-data-directory>

Stream EON video data to a PC

Hardware needed: EON, comma Smays.

You can connect your EON to your pc using the Ethernet cable provided with the comma Smays and you'll be able to stream data from your EON, in real time, with low latency. A useful application is being able to stream the raw video frames at 20fps, as captured by the EON's camera.

Usage:

# ssh into the eon and run loggerd with the flag "--stream". In ../selfdrive/manager.py you can change:
# ...
# "loggerd": ("selfdrive/loggerd", ["./loggerd"]),
# ...
# with:
# ...
# "loggerd": ("selfdrive/loggerd", ["./loggerd", "--stream"]),
# ...

# On the PC:
# To receive frames from the EON and re-publish them. Set PYGAME env variable if you want to display the video stream
python streamer/streamerd.py

Welcomed contributions

• Documentation: code comments, better tutorials, etc..
• Support for other platforms other than Ubuntu 16.04.
• Performance improvements: the tools have been developed on high-performance workstations (12+ logical cores with 32+ GB of RAM), so they are not optimized for running efficiently. For example, ui.py might not be able to run real-time on most PCs.
• More tools: anything that you think might be helpful to others.