Use rsl_comm_py instead
Why rsl_comm_py
repo? Since we support more then UM7
(UM7
, UM8
, shearwater
),
we unified communication code in one repo.
TL;DR: "Swiss army knife" for using
the UM7
orientation sensor with Python 3 (Python 3.6+).
UM7
comes with the
"Serial Software Interface"
for handling / communicating with the sensor, which is currently available for Windows only.
The python
driver provided here is designed to keep you up and running
on different platforms (Linux, Windows, Mac).
If you have the UM7
board and want to use it on Linux (e.g. Ubuntu, Debian, Raspbian, Yocto, Suse, etc.),
Windows or Mac, this repo provides driver code to send / receive individual packets
and receive broadcasts, as well example code how to create a sensor communication object.
In particular, the driver has the following capabilities:
-
read / write single
UM7
registers over UART; -
receive broadcast data from the
UM7
sensor over UART; -
register map and interpretation of the sensor registers.
To get started, you need to know that communicating with the UM7 over
the UART is coded in um7_serial.py
file,
where the UM7Serial
class is defined.
Information about UM7 registers comes to um7_serial.py
from the um7_registers.py
file, where
the accessing to the UM7 registers are stored.
Since it is possible to access the UM7 register map over UART and SPI,
the register data (e.g. addresses, fields, and their meaning) is stored in a separate file.
In the examples
folder we store the examples how to communicate with the
sensor.
The UM7 register description is stored in the SVD file um7.svd
and is parsed by the rsl_svd_parser.py
.
The parser extracts the information from the XML file and keeps it as python data classes.
Below we outline the repo structure:
um7py
: top-level python package;um7py/examples
: package with example code for receiving broadcast / reading / writing UM7 registers;um7py/rsl_xml_svd
: package that stores UM7 registers data in SVD (or System View Description) format and parsing code;um7py/rsl_xml_svd/test
:pytest
tests for SVD parsing;um7py/rsl_xml_svd/RSL-SVD.xsd
: RedShiftLabs SVD XML Schema based on CMSIS5 SVD Schema;um7py/rsl_xml_svd/rsl_svd_parser.py
: parsing code for um7.svd into dataclasses;um7py/rsl_xml_svd/um7.svd
: the UM7 SVD file, which stores register description in xml (in particular SVD format);um7py/templates
: jinja2 templates used for code generation part;um7py/test
: pytest tests for code generation part;um7py/rsl_generate_um7.py
: invoke code generation and save generated results;um7py/rsl_generator.py
: code generation forum7_registers.py
from the SVD file;um7py/serve_um7_autodetect.py
: copies theum7_autodetect.py
script to the desired location;um7py/um7_autodetect.py
: UM7 script for saving configuration for connection to the USB Expansion Board;um7py/um7_broadcast_packets.py
: dataclasses for UM7 broadcast messages;um7py/um7_registers.py
: UM7 register description file;um7py/um7_serial.py
: UM7 UART driver;
UM7 provides serial (UART) and SPI interfaces, hence the two main ways to access the sensor data are UART (serial) or SPI. The differences in short: UART provides broadcast functionality, i.e. when packets can transmitted by the board with a specified frequency (transmission frequencies are set up in configuration registers), and it is possible to issue sensor commands (i.e. accessing command registers). SPI access the sensor register on demand (i.e. no broadcast functionality), and only configuration and data registers can be accessed. Accessing commands is only supported over UART.
When using UM7 over serial, it is possible to connect to the target system (i.e. user's target):
-
to the serial port directly (e.g. when serial pins are wired out as on the Raspberry PI, NVIDIA Jetson Nano, or other board computers with GPIO and UART pins wired out);
-
to the USB port using the USB Expansion Board, which performs USB to serial conversion.
When using the UM7 over SPI, there are also a couple of possibilities:
-
to the SPI pins directly (e.g. Raspberry PI, NVIDIA Jetson Nano), i.e. the pins are wired to the SoC directly;
-
to the USB port using USB to SPI converter, e.g. USB-ISS.
The difference between the two, that in the first case SoC pins support the SPI directly (on the hardware level, which also mirrors in the OS level), then the OS is likely to have the SPI device driver built-in (e.g. Raspberry PI). In the second case, using external converter (e.g. USB-ISS), the device will be shown as a so-called cdc_acm (communication device class), and low-level SPI communication will be done by the converter, yet to the OS the converter will be shown as Abstract Control Model (ACM) USB Device.
pip install um7py
TL;DR: install
(i) pyserial
,
(ii) pyudev
(for Linux),
(iii) dataclasses
(included in standard library since python3.7
, needs to be installed for 3.6
).
If you want to use SPI: if using on Linux and use SPI bus directly, install spidev
,
otherwise if using USB-ISS install usb_iss
python package.
Create UM7
serial communication object, UM7 connected to a port /dev/ttyUSB0
,
and read the firmware version:
from um7py import UM7Serial
um7_serial = UM7Serial(port_name='/dev/ttyUSB0')
print(f"um7 firmware revision: {um7_serial.get_fw_revision}")
Reading all types of broadcast packets from UM7
, 1000 packets in total:
from um7py import UM7Serial
um7_serial = UM7Serial(port_name='/dev/ttyUSB0')
for packet in um7_serial.recv_broadcast(num_packets=1000):
print(f"packet: {packet}")
Reading the raw sensor data broadcast packets from UM7
, not limiting number of packets:
from um7py import UM7Serial
um7_serial = UM7Serial(port_name='/dev/ttyUSB0')
for packet in um7_serial.recv_all_raw_broadcast():
print(f"packet: {packet}")
Reading 100 processed sensor data broadcast packets from UM7
:
from um7py import UM7Serial
um7_serial = UM7Serial(port_name='/dev/ttyUSB0')
for packet in um7_serial.recv_all_proc_broadcast(num_packets=100):
print(f"packet: {packet}")
Reading the Euler angles broadcast packets from UM7
:
from um7py import UM7Serial
um7_serial = UM7Serial(port_name='/dev/ttyUSB0')
for packet in um7_serial.recv_euler_broadcast():
print(f"packet: {packet}")
Reading the CREG_COM_SETTINGS
configuration register from UM7
:
from um7py import UM7Serial
um7_serial = UM7Serial(port_name='/dev/ttyUSB0')
print(f"received value: {um7_serial.creg_com_settings}")
Writing 40 (changing ALL_RAW_RATE
to 40 Hz) to the CREG_COM_RATES2
register:
from um7py import UM7Serial
um7_serial = UM7Serial(port_name='/dev/ttyUSB0')
um7_serial.creg_com_rates2 = 40
Take a look at the available examples.
In order to use the python
driver functionality one first needs to
create a communication object (in our case UM7Serial
or UM7SPI
).
The construction of the UART communication object can be done
either by specifying a port_name
directly (e.g. /dev/ttyS0
), or
by specifying the device
file that stores USB2Serial config
(e.g. um7_A500CNP8.json
). The device
argument shall
only be used, when UM7 is connected via the
USB Expansion Board,
the device
stores properties of the expansion board. Why?
The issue to keep in mind that when the sensor is re-plugged,
it might appear to the OS as different serial connection,
i.e. when first plugged in the OS detects the device as
/dev/ttyS0
, then after re-plugging exactly the same
sensor might appear by different port name, e.g. /dev/ttyS1
, which
means the user code needs to be changed.
If using the device
, we store properties of the
USB Expansion Board
in a JSON file (e.g. converter chip ID)
and search connection which match with the properties, and
in this case connecting to the sensor, even if is shown as a
different serial connection by the OS.
So the communication object can either be created with specifying the port_name
:
from um7py import UM7Serial
um7_serial = UM7Serial(port_name='/dev/ttyUSB0')
Or with specifying the device
:
from um7py import UM7Serial
um7 = UM7Serial(device='um7_A500CNP8.json')
The two options are exclusive, i.e. specifying both port_name
and device
will not work.
Accessing to the individual registers is done via python
properties.
Properties for register names are all lower-case, split by _
.
For example, reading the CREG_COM_RATES1
:
from um7py import UM7Serial
um7 = UM7Serial(device='um7_A500CNP8.json')
um7.creg_com_rates1
Note, that reading single register is quite a slow operation, since one first constructs and sends a packet, and then parses output for response. Reading single registers is not recommended for reading sensor data, since it might happen, that data from different sensor registers come from different measurements. We strongly advice to use broadcast messages for reading sensor and fusion data.
UM7
sends different types of broadcast messages over the UART.
These messages are e.g. HEALTH packet (i.e. the DREG_HEALTH
register),
raw sensor data (raw gyro, accelerometer, and magnetometer, and temperature),
processed sensor data (processed gyro, accelerometer, and magnetometer),
Euler angles, quaternions.
These data packets are stored in the repo as dataclasses in the file um7_broadcast_packets.py. Note, that only payload stored in the dataclasses, and all the checks (e.g. checksum, data length) is done during broadcast reception.
For example, the raw data broadcast message has the following payload:
from dataclasses import dataclass
@dataclass
class UM7AllRawPacket:
gyro_raw_x: int
gyro_raw_y: int
gyro_raw_z: int
gyro_raw_time: float
accel_raw_x: int
accel_raw_y: int
accel_raw_z: int
accel_raw_time: float
mag_raw_x: int
mag_raw_y: int
mag_raw_z: int
mag_raw_time: float
temperature: float
temperature_time: float
We are so grateful for the open source community for creating open source UM7 driver versions and sharing it with a world! We are inspired by your work, and at the same time want to improve: provide UART and SPI communication, in detail documentation and explanations to facilitate the start for new users.
The acknowledgments go to:
-
Daniel Kurek and his um7 repository, for implementing the first driver for interfacing with UM7;
-
Till Busch and his um7 fork of Daniel's Kurek repo, for extending on the Daniel's work and adding new functionality.
Dr. Konstantin Selyunin, for suggestions / questions / comments please contact selyunin [dot] k [dot] v [at] gmail [dot] com.