Using Pymavlink Libraries (mavgen)

Pymavlink is a low level and general purpose MAVLink message processing library, written in Python. It has been used to implement MAVLink communications in many types of MAVLink systems, including a GCS (MAVProxy), Developer APIs (DroneKit) and numerous companion computer MAVLink applications.

The library can be used with Python 3.5+ and supports both MAVLink 1 and MAVLink 2 versions of the protocol.

This topic explains how to get and use the Pymavlink MAVLink Python libraries (generated using mavgen).

Pymavlink is developed in its own project, which includes the command line MAVLink generator (mavgen), Python bindings to create Pymavlink, and other useful tools and utilities. MAVLink includes the Pymavlink repository as a submodule. While you can work with that project directly, this documentation explains how to work with pymavlink using the MAVLink project.

If you are writing a MAVLink application to communicate with an autopilot you may prefer to use a higher level library like MAVSDK-Python or DroneKit-Python. These implement a number of MAVLink microservices.

The following are instructions for how to obtain the MAVLink libraries for the standard and custom dialects. If the standard dialect is to be used, the instructions are much simpler. For custom dialects the library needs to be generated first.

If you need a standard dialect then you can install these (for both MAVLink 1 and 2) with pymavlink using pip:

pip install pymavlink

The PyPi repository takes message definitions from the ArduPilot/mavlink fork, which may diverge slightly from MAVLink/mavlink.

If you need libraries for a custom dialect then you will need to install the code generator mavgen and generate the libraries yourself. You will also need to include them in pymavlink and install them locally on your system.

  1. Generate the Python MAVLink libraries for your custom dialect.
  2. Copy the generated .py MAVLink dialect library file(s) into the appropriate directory of your clone of the mavlink repository:
    • MAVLink 2: pymavlink/dialects/v20
    • MAVLink 1: pymavlink/dialects/v10
  3. Open a command prompt and navigate to the pymavlink directory.
  4. If needed, uninstall previous versions:

    pip uninstall pymavlink
  5. Install dependencies if you have not previously installed pymavlink using pip:

    python3 -m pip install -r pymavlink/requirements.txt
  6. Run the python setup program:

    python install --user

The generated MAVLink libraries can then be used in the same way as those installed using pip.


The pymavlink package includes the dialect-specific generated modules, which provide low-level functionality to encode and decode messages, and apply and check signatures.

Generally speaking, most developers will use the mavutil module to set up and manage the communication channel, because it makes getting started very easy. This module provides simple mechanisms for setting up links, sending and receiving messages, and querying some basic autopilot properties such as the currently active flight mode for example. It provides access to the dialect module used for encoding, decoding and signing messages via an attribute (mav).

There are several main caveats to be aware of when using mavutil:

  • The link does not properly handle multiple systems running on the same port. If you need a multi-vehicle network see source-system-filtering.
  • The module is optimised for ArduPilot and some functions may not work properly on other autopilots.
  • mavutil is still a relatively low-level MAVLink API. It has limited support for even the most common MAVLink microservices.

You don't have to use mavutil but it includes a lot of useful code that you might otherwise have to write from scratch. In particular the connection code and methods to filter incoming messages are useful for any autopilot.

The set of modules in the pymavlink package are listed below:

  • \dialects\v20\* and \dialects\v10\*: Dialect modules corresponding to each source XML message definition for MAVLink v2 and v1, respectively. Each dialect module contains:
    • constants for all enums and enum values defined in the XML file.
    • a set of constants for message identifiers.
    • a class for each type of MAVLink message defined in the XML file.
    • a MAVLink class, which can be used to send and receive messages:
      • has _send and _decode functions for each message type.
      • has methods to check and apply signatures.
      • has lower-level methods for packing and parsing data.
  • mavutil: MAVLink utility functions for setting up communication links, receiving and decoding messages, running periodic tasks, etc.
    • mavutil.mavlink_connection(device, baud, ...) for setting up a link to (initially) listen for messages or send messages on a channel (e.g. udp, serial, etc.). This returns an object representing the connection. You can use:
      • themav attribute for accessing the selected dialect/protocol module to encode and send messages
      • setup_signing() for setting up signing
      • recv_match() for capturing messages with particular names or field values
    • The connection allows you to do a lot of other useful work: get all autopilot parameters, access last message of each type received, get the autopilot telemetry including current flight mode or armed state, etc.
  • mavwp: Load/save waypoints, geofence, rally points.
  • mavparm: Load/save sets of MAVLink parameters.
  • mavextra: Useful functions for converting values and messages (e.g. metres/second to Km/h, eulers in radians from quaternion etc.).
  • mavexpression (internal): MAVLink expression evaluation functions.

Choosing the Dialect/MAVLink Version

Choosing the Dialect/MAVLink version depends on whether you are using mavutil for link management or working directly with dialect files.

By default mavutil sets up the link to use the MAVLink 1 ardupilotmega dialect for sending/receiving. You can change this by setting environment variables:

  • MAVLINK_DIALECT: Set to string name for the dialect file (without XML extension).
  • MAVLINK20: Set to 1 (if unset then default to MAVLink 1)
  • MDEF: Location of message definition libraries

You can also change the dialect by passing its name to mavutil.mavlink_connection() when setting up a connection.

If you are not using mavutil then you can import the dialect file that you want to use directly:

# Import ardupilotmega module for MAVLink 1
from pymavlink.dialects.v10 import ardupilotmega as mavlink1

# Import common module for MAVLink 2
from pymavlink.dialects.v20 import common as mavlink2

Setting up a Connection

The mavutil module provides the mavlink_connection() method for setting up communication links to MAVLink systems over serial ports, tcp, or udp channels. It can also connect to a file object, which is useful when working with telemetry logs.

The method returns an object that represents a single system, but will collect messages from multiple systems on the link. This is OK for two-system networks, but if you need to connect over a multi-vehicle IP network see source-system-filtering.

The mavlink_connection() method takes a connection string defining the channel, and some optional arguments for setting the baud rate and other properties of the link (the format of the connection string automatically sets, and may override, some optional arguments).

For example, to connect to the standard MAVLink simulator UDP port, and wait for HEARTBEAT messages:

from pymavlink import mavutil

# Start a connection listening on a UDP port
the_connection = mavutil.mavlink_connection('udpin:localhost:14540')

# Wait for the first heartbeat
#   This sets the system and component ID of remote system for the link
print("Heartbeat from system (system %u component %u)" % (the_connection.target_system, the_connection.target_component))

# Once connected, use 'the_connection' to get and send messages

The udpin prefix used above creates a socket to listen for a UDP connection on the specified port. This is the normal way to connect an autopilot simulator). The complementary udpout prefix creates a socket that initiates an IP connection:

the_connection = mavutil.mavlink_connection('udpout:localhost:14540')

Generally the API selects a sensible baud rate for the connection type. Other mavlink_connection() parameters you may wish to change include: source_system (default 255), source_component (default 0) and dialect (default ArduPilot).

Connection Strings

The mavutil.mavlink_connection() connection string has the format:



  • protocol (optional): The IP protocol. If not specified pymavlink will attempt to determine if the address is a serial port (e.g. USB) or a file, and if not will default to a UDP address.
    • tcp: Initiate a TCP connection on the specified address and port.
    • tcpin: Listen for a TCP connection on the specified address and port.
    • udpin: Listen for a UDP connection on the specified address and port.
    • udpout: Initiate a TCP connection on the specified address and port.
    • udp: By default, same as udpin. Set mavlink_connection parameter input=False to make same as udpout.
    • udpcast: Broadcast UDP address and port. This is the same as udp with mavlink_connection() parameters input=False and broadcast=True.
  • address: IP address, serial port name, or file name
  • port: IP port (only if address is an IP address)

Some of the strings you can use for different types of connections are listed below.

Connection typeConnection string
Linux computer connected to the vehicle via USB/dev/ttyUSB0
Linux computer connected to the vehicle via serial port (RaspberryPi example)/dev/ttyAMA0 (also set baud=57600)
MAVLink API listening for SITL connection via UDPudpin:localhost:14540 (or udp:localhost:14540,,etc.)
MAVLink API initiating a connection to SITL via UDPudpout:localhost:14540 (or udpout:
GCS connected to the vehicle via UDP127.0.0.1:14550 or udp:localhost:14550
SITL connected to the vehicle via TCPtcp: (ArduPilot only, PX4 does not support TCP)
OSX computer connected to the vehicle via USBdev/cu.usbmodem1
Windows computer connected to the vehicle via USB (in this case on COM14)com14
Windows computer connected to the vehicle using a 3DR Telemetry Radio on COM14com14 (also set baud=57600)

While MAVLink does not define the UDP ports used for different purposes, there is a defacto standard that MAVLink APIs should listen for SITL connections on UDP port 14540 while a GCS should listen for connections on UDP 14550.

Sending Messages

MAVLink is the main protocol handling class. It is defined in each dialect module, and includes a <message_name>_send() method for all messages in the dialect's message definition.

The message field values are passed as arguments to the function. Fields that are the same for all messages are defined in the class - e.g. source system, source component. Each message is documented in the dialect source code, even when it was automatically generated.

For example, the system_time_send() function is used to send the SYSTEM_TIME message as shown below:

def system_time_send(self, time_unix_usec, time_boot_ms, force_mavlink1=False):
    The system time is the time of the master clock, typically the
    computer clock of the main onboard computer.

    time_unix_usec    : Timestamp (UNIX epoch time). (uint64_t)
    time_boot_ms      : Timestamp (time since system boot). (uint32_t)

If you are using mavutil for link management then the mav attribute provides access to a configured MAVLink class object that you can use for sending messages. For example, to send the SYSTEM_TIME message using a link named the_connection:

the_connection.mav.system_time_send(time_unix_usec, time_boot_ms)

Other examples can be seen in Publishing a Heartbeat and Requesting Specific Messages below.

If you are not using mavutil you will need to create and set up the MAVLink object yourself so it knows about the channel that it should use for sending messages, which is represented by the file attribute.

Receiving Messages

If you just want to synchronously access the last message of a particular type that was received, and when it was received, you can do so using the connection's mavutil.messages dictionary. For example, if you are using a mavutil link named the_connection you can do:

    altitude = the_connection.messages['GPS_RAW_INT'].alt  # Note, you can access message fields as attributes!
    timestamp = the_connection.time_since('GPS_RAW_INT')
    print('No GPS_RAW_INT message received')

Alternatively you can use the mavutil recv_match() method to wait for and intercept messages as they arrive:

def recv_match(self, condition=None, type=None, blocking=False, timeout=None):
    '''Receive the next MAVLink message that matches the given type and condition
    type:        Message name(s) as a string or list of strings - e.g. 'SYS_STATUS'
    condition:   Condition based on message values - e.g. 'SYS_STATUS.mode==2 and SYS_STATUS.nav_mode==4'
    blocking:    Set to wait until message arrives before method completes.
    timeout:     ? <!-- timeout for blocking message when the system will return. Is this just a time? -->

For example using the_connection set up as before, you can wait for any message as shown:

msg = the_connection.recv_match(blocking=True)

If you instead want to just get a particular message with certain attribute values you might instead do:

# Wait for a 'SYS_STATUS' message with the specified values.
msg = the_connection.recv_match(type='SYS_STATUS', condition='SYS_STATUS.mode==2 and SYS_STATUS.nav_mode==4', blocking=True)

You should also check that the message is valid before attempting to use it:

msg = m.recv_match(type='SYS_STATUS',blocking=True)
if not msg:
if msg.get_type() == "BAD_DATA":
    if mavutil.all_printable(
    #Message is valid
    # Use the attribute
    print('Mode: %s' % msg.mode)

The returned object is the subclass of MAVLink_message for the specific message. You can access the message fields as class attributes as shown for the mode in the above code fragment. If needed, you can query MAVLink_message for information about the signature, CRC and other header information.

Requesting Specific Messages

A remote system will typically stream a default set of messages to a connected GCS, camera or other system. This set may be hard coded, and is necessarily limited to reduce traffic on the channel.

A system can request that additional messages are sent as a stream, or change the rate at which existing streamed messages are sent, using the MAV_CMD_SET_MESSAGE_INTERVAL command. This can be sent in either a COMMAND_LONG or COMMAND_INT, if supported by the flight stack.

For more information see How to request messages

Publishing a Heartbeat

All MAVLink components should periodically broadcast a HEARTBEAT message and listen for heartbeats from other systems. Systems consider themselves connected to another system as long as they regularly receive a HEARTBEAT from it.

Sending Messages explains how messages are sent.

The HEARTBEAT message can be sent using MAVLink.heartbeat_send() message in the generated Python dialect file. The method definition is provided below:

def heartbeat_send(self, type, autopilot, base_mode, custom_mode, system_status, mavlink_version=3, force_mavlink1=False):
    The heartbeat message shows that a system is present and responding.
    The type of the MAV and Autopilot hardware allow the
    receiving system to treat further messages from this
    system appropriate (e.g. by laying out the user
    interface based on the autopilot).

    type              : Type of the MAV (quadrotor, helicopter, etc.) (type:uint8_t, values:MAV_TYPE)
    autopilot         : Autopilot type / class. (type:uint8_t, values:MAV_AUTOPILOT)
    base_mode         : System mode bitmap. (type:uint8_t, values:MAV_MODE_FLAG)
    custom_mode       : A bitfield for use for autopilot-specific flags (type:uint32_t)
    system_status     : System status flag. (type:uint8_t, values:MAV_STATE)
    mavlink_version   : MAVLink version, not writable by user, gets added by protocol because of magic data type: uint8_t_mavlink_version (type:uint8_t)

Assuming you are using a mavutil link named the_connection, which is returned by mavutil.mavlink_connection(), you can send a heartbeat as follows:

# Send heartbeat from a GCS (types are define as enum in the dialect file).
                                                mavutil.mavlink.MAV_AUTOPILOT_INVALID, 0, 0, 0)

# Send heartbeat from a MAVLink application.
                                                mavutil.mavlink.MAV_AUTOPILOT_INVALID, 0, 0, 0)

The various types used above come from enum in the dialect file.

The rate at which the heartbeat should be sent depends on the channel, but is normally 1Hz.

Generally it should be sent from the same thread as all other messages. This is in order to ensure that the heartbeat is only published when the thread is healthy.

Message Signing

Pymavlink supports Message Signing (authentication) when using MAVLink 2.

The Pymavlink library already implements almost all of the expected behaviour for signing messages. All you need to do is provide a secret key and initial timestamp, optionally specify whether or not outgoing messages should be signed, a link id, and a callback for determining which unsigned messages (if any) will be accepted.

For more information see Message Signing (Pymavlink).


See Examples (pymavlink)


Pymavlink questions can be raised in the normal MAVLink support channels.

The Pymavlink Gitter channel also has an active support community.

results matching ""

    No results matching ""