valgur/velodyne_decoder

Fully-featured C++/Python Velodyne packet decoder

velodyne_decoder's Issues

Time

How to get the proper timestamp using this package?
By default, the output comes out as negative in float32. Can it be converted to long?

Request for Apple Silicon Support

Hi valgur,
Thank you for providing such an amazing tool. I moved my work flow from an intel MacBook Pro to the Silicon environment, the installation cannot work, tried both pip install velodyne-decoder (cannot install) and pip install git+https://github.com/valgur/velodyne_decoder.git (says installed successfully, but the vd.Config(model="Alpha Prime", rpm=600) cannot be created.

Error Message:
    config = vd.Config(model="Alpha Prime", rpm=600)
TypeError: __init__(): incompatible constructor arguments. The following argument types are supported:
    1. velodyne_decoder.velodyne_decoder_pylib.Config(*, model: Optional[velodyne_decoder::ModelId] = None, calibration: Optional[velodyne_decoder::Calibration] = None, single_return_mode_info: bool = False, min_range: float = 0.1, max_range: float = 200, min_angle: float = 0, max_angle: float = 360, cut_angle: Optional[float] = None, timestamp_first_packet: bool = False, use_device_time: bool = False)

Invoked with: kwargs: model='Alpha Prime', rpm=600

Thanks again,
Crear

How About Direct Connection to LiDAR and read the PCAP Stream from Static IP address without Using ROS?

Now we have successfully connect our Jetson Orin to the LiDAR device and we can use tcpdump to generate .pcap files without involving ROS. (.pcap files are just the network traffic packages) However, to test real-time processing, we would like to directly analyze the LiDAR stream instead of saving and then analyzing the pcap files. May I ask how to directly use the IP address and maybe port number as LiDAR input?

Unexpected Attribute Error: 'bytes' object has no attribute 'data'

Hi Martin,
Thanks for creating this decoder and Happy New Year!
I'm using the velodyne_decoder to decode a 5-min-long pcap file from an Alpha Prime, which can be opened and played using Veloview from the first frame to the last frame.
However, while using the velodyne_decoder to open the same file, it exits at frame 234/3000, and throws this error while trying to use the "next(The generator object read_pcap)" in the while loop:

File ~\anaconda3\envs\LiDAR\lib\site-packages\velodyne_decoder_init_.py:42 in read_pcap
data = dpkt.ethernet.Ethernet(buf).data.data.data
AttributeError: 'bytes' object has no attribute 'data'

I suspect that the Veloview has some exception handling procedure to locate the next correct frame, because it feels like there are some disturbance in the pcap file (maybe happened during ethernet transmission) that interrupt the decoding. Do you have any suggestion for this?

Thanks,
Yi

RuntimeError: Unable to cast Python instance to C++ type (compile in debug mode for details)

Environment:

• ROS version: melodic
• Python: 2.7
• Velodyne type: VLP16

Issue
The default example in the README cause the following error:

    cloud_arrays.append(decoder.decode_message(scan_msg))
RuntimeError: Unable to cast Python instance to C++ type (compile in debug mode for details)

After installing the package in debug, I got

    cloud_arrays.append(decoder.decode_message(scan_msg))
RuntimeError: Unable to cast Python instance of type <type 'str'> to C++ type 'std::array<unsigned char, 1206ul>'

Workaround
To make the demo work (without changing the package), I had to convert each packet's data from a str type to a bytearray type

for packet in scan_msg.packets:
  packet.data = bytearray(packet.data)

By the way, thanks for sharing this package! Very usefull :)

Decoding Velodyne data from a PCAP file have wrong number of pcd frames for VLP-32C

I'm using our PCAP file, on Linux and MacOS, in duration of 20 sec, created by Velodyne VLP-32C, with 600 rpm, 10Hz.
In VeloView I get as expected 200 pcd frames with this configuration, but out of velodyne_decoder I get 400 frames.

By testing velodyne_decoder on Velodyne HDL32E with 600 rpm, 10 Hz on following dataset I get 843 pcd frames, in VeloView I get 2013?
Kitaware dataset: 2014-11-10-10-36-54_Velodyne-VLP_10Hz-County Fair.pcap, from https://data.kitware.com/api/v1/item/5b7fff608d777f06857cb53a/download

Is it possible that the calculation of frames per rotation and second is not right?

error on import in develop branch

After installing the package from develop branch, I am getting the following error on import.

`---------------------------------------------------------------------------
ModuleNotFoundError Traceback (most recent call last)
in
----> 1 import velodyne_decoder as vd

/opt/conda/lib/python3.8/site-packages/velodyne_decoder/init.py in
2 from collections import namedtuple
3
----> 4 from velodyne_decoder.velodyne_decoder_pylib import *
5 from velodyne_decoder.velodyne_decoder_pylib import version as _pylib_version
6

ModuleNotFoundError: No module named 'velodyne_decoder.velodyne_decoder_pylib'`

v3.0.0 release

With the significantly simpler configuration, dual-return mode support, reworked scan batching and full precise-timing support for all model versions, I am using this opportunity to create a backwards-incompatible v3.0.0 release soon.

These changes reside in the develop branch in the mean time.

Some ideas and loose ends I would still like to implement before the release:

• autodetection of HDL-64E from input file in Python
• gps_time→use_device_time
• Move __init__.py contents into separate Python files.
• StreamDecoder →ScanBatcher
• Return both arrival and device timestamps for scans, drop use_device_time param
• Always use packet stamp for within-scan timestamps
• Add getters for model ID and return mode to PacketDecoder
• ros-drivers/velodyne#518
• Make return_type a separate field
• Add column field
• Fix __version__ info
• Unit tests for PCAP decoding by model
• PCAP → rosbag conversion
• Support for rosbags library input (#6)
• Support for re-batching of ROS VelodyneScan packets.
• HDL-64E calib extraction from rosbag
• HDL-64E autodetection for rosbag
• Decode by ROS topic
• TelemetryPacket output support for PCAP
• VLS-128 timings validation
• Update README
• Update ChangeLog
• Sphinx documentation

Feature Request: Reading Raw Azimuth Angle, Laser ID, Distance, and Timestamp

Hi Martin,
Recently we see increasing research with brilliant ideas of using the raw spherical coordinate system data without translating to the cartesian coordinate system. I'm wondering if it's possible to allow passing a parameter to the vd.read_pcap function like: vd.read_pcap(pcap_file_name, config, spherical_coordinate_system=True) to directly return the spherical coordinates like azimuth angle, laser Id, distance? This would save us tons of efforts in reproducing the methodologies proposed in those papers.

Thanks in advance,
Crear

Inconsistent frame sizes

When processing a VLP-16 .PCAP file, the output frames are inconsistent in size. One frame will contain ~32,000 points and the next ~8,000 points. There is a consistent pattern of large, then small frames.

Using VeloView on the same data file, and exporting .CSV files produce frames of ~19,000 points in all frames.

Broken VLS-128 config

While testing the VLS-128 integration, the reading of the rosbag failed with:

File "*****/python3.6/site-packages/velodyne_decoder/__init__.py", line 94, in read_bag decoder = ScanDecoder(config) RuntimeError: Unable to open calibration file: ******/python3.6/site-packages/velodyne_decoder/calibrations/VLS-128.yml

This is because the config file within the repo is broken and some of the laser id's have missing data. Is it possible to fix this issue?

decoded data arrangement

I used the snippet you provided for Decoding Velodyne data from a PCAP file , however the data comes out as
Y , negative X , Z
instead of x,y,z . I compared it to veloview extracted csv and it indeed came out as
Points_m_XYZ:1,-Points_m_XYZ:0,Points_m_XYZ:2

I know I can just multiply by -1 and rearrange , but am I doing something wrong ?

Sensor type : VLP-16