6.4. Performance & Optimization

6.4.1. Ob_benchmark tool

The goal of this tool is to benchmark the performance of various OrbbecSDK_ROS2 camera configurations. The benchmark results depend on the camera and settings used.(Currently only works with ROS2 Humble)

You can find example usage code in the example.

6.4.1.1. Tool Configuration (start_benchmark_params.json)

{
    "start_benchmark_params": {
        "camera_name": [
            "camera_01",
            "camera_02",
            "camera_03",
            "camera_04"
        ],
        "process_name": "component_conta",
        "switch_cycle": 300,
        "test_cycle": 1,
        "skip_number": 30
    }
}

  • camera_name: Names of the cameras to be configured. Example: "camera_01", "camera_02", etc.

  • process_name: The name of the process to be monitored. For example, "component_conta" will monitor the data of the container process.

  • switch_cycle: The cycle time for switching configurations, in seconds. For example, setting it to 300 means the configuration will switch every 300 seconds.

  • test_cycle: The testing cycle, in seconds. For example, setting it to 1 means the tool will collect data for the monitored process every 1 second.

  • skip_number: The number of data points to skip. For example, setting it to 30 means that the first 30 data points will be ignored.

6.4.1.2. Camera configuration (launch files)

In the launch folder, there are multiple.launch.py files (ob_benchmark_0.launch.py, ob_benchmark_1.launch.py, …, ob_benchmark_19.launch.py). Each file corresponds to a different camera configuration.

6.4.1.3. Running the ob_benchmark tool

To run the tool, use the following commands:

source install/setup.bash
ros2 run orbbec_camera ob_benchmark_node

6.4.1.4. Output Data Files

The output data files will be stored in the ob_benchmark folder with filenames like 0.csv, 1.csv, …, 19.csv. For example:

  • 0.csv contains data from the ob_benchmark_0.launch.py configuration.

  • 1.csv contains data from the ob_benchmark_1.launch.py configuration.

6.4.2. Reducing CPU Usage with Orbbec ROS Package

You can find example usage code in the example.

This document outlines strategies for minimizing CPU usage in the OrbbecSDK_ROS2 v2 environment when using Gemini 330 series cameras. The firmware version must be no lower than 1.4.10, and device should be set to Default.

6.4.2.2. Launch Files Used for Testing

  • gemini_330_series_lower_cpu_usage.launch.py

  • multi_camera_lower_cpu_usage.launch.py

6.4.2.3. Test environment

Hardware Configuration

  • CPU: Intel i7-8700 @ 3.20GHz

  • Memory: 24 GB

  • Storage: Micron 2200S NVMe 256GB

  • GPU: NVIDIA GeForce GTX 1660Ti

  • OS: Ubuntu22.04

ROS Configuration

  • ROS Version: ROS2 Humble

  • SDK Version: OrbbecSDK_ROS2 v2.2.1

Camera Setup

  • Devices: 2x Gemini 335, 1x Gemini 336, 1x Gemini 336L

  • Firmware Version: 1.4.10

6.4.2.4. Test Setup

Stream Settings:

  • Depth / IR Left / IR Right: 848×480 @ 30fps

  • Color: 848×480 @ 30fps

Note: The following CPU usage data focuses on uvc_backend, color_format and various filter combinations.

6.4.2.5. Test Results

uvc_backend Comparison (RGB format)

libuvc CPU Usage v4l2 CPU Usage Absolute Change
182.8% 118.8% -64.0%

The CPU usage can be significantly reduced with v4l2 backend. In our implementation, v4l2 works without requiring any patches to the Linux kernel, allowing users to easily switch between v4l2 and libuvc and maintaining full compatibility with standard Linux distributions.

color_format Comparison (MJPG vs RGB)

Backend MJPG CPU Usage RGB CPU Usage Absolute Change
libuvc 347.7% 182.8% -164.9%
v4l2 170.0% 118.8% -51.2%

The CPU usage can be reduced if the RGB format is selected instead of MJPG, since the decoding of MJPG image will consume the host CPU resource.

Filter Configuration Impact

Filters Applied libuvc CPU Usage CPU Usage Increase v4l2 CPU Usage CPU Usage Increase
No Filter (benchmark) 182.8% 0.0%(benchmark) 118.8% 0.0%(benchmark)
(software)noise_removal_filter 218.0% +35.2% 128.5% +9.7%
(software)noise_removal_filter + spatial_filter 469.6% +286.8% 336.7% +217.9%
hardware_noise_removal_filter 186.3% +3.5% 115.4% -3.4%
hardware_noise_removal_filter + spatial_filter 251.3% +68.5% 152.5% +33.7%

Based on the test results, using only the hardware_noise_removal_filter results in a negligible change in CPU usage for both libuvc (+3.5%) and v4l2 (-3.4%) compared to the no-filter benchmark, as this filter runs internally on the camera hardware. In contrast, other filters execute on the host system. Adding the spatial_filter to the hardware filter leads to a moderate increase in CPU usage, while applying the software-based noise_removal_filter —either alone or combined with spatial_filter —significantly increases CPU load. To maintain low CPU usage, it is recommended to avoid software-based filters and rely solely on the hardware_noise_removal_filter.

6.4.2.6. Further Optimizationa

Parameter Recommendation Note
depth_registration false or true with align_mode=HW Software alignment consumes more CPU
enable_point_cloud false Disabling point cloud reduces CPU usage
enable_colored_point_cloud false Disabling colored point cloud reduces CPU usage

6.4.3. Efficient intra-process communication

6.4.3.1. Introduction

Our ROS2 Wrapper node supports zero-copy communications if loaded in the same process as a subscriber node. This can reduce copy times on image/pointcloud topics, especially with big frame resolutions and high FPS.

You will need to launch a component container and launch our node as a component together with other component nodes. Further details on “Composing multiple nodes in a single process” can be found here.

Further details on efficient intra-process communication can be found here.

6.4.3.2. Example

Manually loading multiple components into the same process

  • Start the component:

    ros2 run rclcpp_components component_container

  • Add the wrapper:

    ros2 component load /ComponentManager orbbec_camera orbbec_camera::OBCameraNodeDriver -e use_intra_process_comms:=true

    Load other component nodes (consumers of the wrapper topics) in the same way.

Using a launch file

ros2 launch orbbec_camera gemini_intra_process_demo_launch.py

Limitations

  • Node components are currently not supported on RCLPY

  • Compressed images using image_transport will be disabled as this isn’t supported with intra-process communication

6.4.4. Fast DDS Optimization for Orbbec Camera with ROS2

When operating with the default configuration, Fast DDS exhibits suboptimal transmission efficiency, resulting in significant image transmission delays when used with the Orbbec camera in ROS2. This document provides guidance on optimizing Fast DDS to enhance image transfer efficiency.

6.4.4.1. Adjusting System Parameters

IP Fragmentation Time

  • Path: /proc/sys/net/ipv4/ipfrag_time (default: 30 seconds)

  • Purpose: Defines the duration that IP fragments are kept in memory.

  • Adjustment: Decrease this value to reduce the time window where no fragments are received, which can help reduce delays. Consider the specific needs of your environment as this setting affects all incoming fragments.

    Example: Set to 3 seconds.

    sudo sysctl net.ipv4.ipfrag_time=3

IP Fragmentation Memory Threshold

  • Path: /proc/sys/net/ipv4/ipfrag_high_thresh (default: 262144 bytes)

  • Purpose: Sets the maximum memory used to reassemble IP fragments.

  • Adjustment: Increase this value to allow more memory for fragment reassembly, which can improve handling of larger data packets.

    Example: Increase to 128 MB.

    sudo sysctl net.ipv4.ipfrag_high_thresh=134217728

Maximum Buffer Sizes

  • Purpose: Configures the maximum buffer sizes for receiving and sending data, which is critical for high-throughput data transmission.

  • Adjustment: Set the maximum buffer sizes for both receiving and sending operations.

    Commands:

    sudo sysctl -w net.core.rmem_max=2147483647
sudo sysctl -w net.core.rmem_default=2147483647
sudo sysctl -w net.core.wmem_max=2147483647
sudo sysctl -w net.core.wmem_default=2147483647

Alternatively, make these settings permanent by adding them to the /etc/sysctl.d/10-fastrtps-max.conf file.

sudo gedit /etc/sysctl.d/10-fastrtps-max.conf

add blow lines to the file:

net.core.rmem_max=2147483647
net.core.rmem_default=2147483647
net.core.wmem_max=2147483647
net.core.wmem_default=2147483647

then save and exit the file. run sudo sysctl -p to apply the changes.

For detailed guidance, refer to ROS 2 DDS Tuning Documentation.

6.4.4.2. Fast DDS Configuration

Below is an example of a Fast DDS configuration file optimized for ROS2 usage with the Orbbec camera. This configuration enhances the overall data transmission by adjusting buffer sizes and transport settings.

Configuration File: shm_fastdds.xml

Place this file in the $HOME directory.

<?xml version="1.0" encoding="UTF-8"?>
<profiles xmlns="http://www.eprosima.com/XMLSchemas/fastRTPS_Profiles">
    <transport_descriptors>
        <transport_descriptor>
            <transport_id>UDP_transport</transport_id>
            <type>UDPv4</type>
            <maxInitialPeersRange>10</maxInitialPeersRange>
            <maxMessageSize>65000</maxMessageSize>
            <sendBufferSize>1048576</sendBufferSize>
            <receiveBufferSize>1048576</receiveBufferSize>
        </transport_descriptor>
    </transport_descriptors>
    <participant profile_name="participant_profile_ros2" is_default_profile="true">
        <rtps>
            <name>profile_for_ros2_context</name>
            <userTransports>
                <transport_id>UDP_transport</transport_id>
            </userTransports>
            <useBuiltinTransports>false</useBuiltinTransports>
            <sendSocketBufferSize>1048576</sendSocketBufferSize>
            <listenSocketBufferSize>1048576</listenSocketBufferSize>
            <builtin>
                <initialPeersList>
                    <locator>
                        <udpv4>
                            <address>127.0.0.1</address>
                        </udpv4>
                    </locator>
                </initialPeersList>
            </builtin>
        </rtps>
    </participant>
    <data_writer profile_name="default publisher profile" is_default_profile="true">
        <qos>
            <publishMode>
                <kind>ASYNCHRONOUS</kind>
            </publishMode>
            <latencyBudget>
                <duration>
                    <sec>0</sec>
                    <nanosec>1000000</nanosec>
                </duration>
            </latencyBudget>
        </qos>
        <historyMemoryPolicy>PREALLOCATED_WITH_REALLOC</historyMemoryPolicy>
    </data_writer>
    <data_reader profile_name="default subscription profile" is_default_profile="true">
        <qos>
            <data_sharing>
                <kind>AUTOMATIC</kind>
            </data_sharing>
            <latencyBudget>
                <duration>
                    <sec>0</sec>
                    <nanosec>1000000</nanosec>
                </duration>
            </latencyBudget>
        </qos>
        <historyMemoryPolicy>PREALLOCATED_WITH_REALLOC</historyMemoryPolicy>
    </data_reader>
</profiles>

Environment Variables

Set the following environment variables to use the custom Fast DDS profile:

export RMW_IMPLEMENTATION=rmw_fastrtps_cpp
export FASTRTPS_DEFAULT_PROFILES_FILE=$HOME/shm_fastdds.xml
export RMW_FASTRTPS_USE_QOS_FROM_XML=1

This configuration aims to optimize the data flow and reduce transmission delays, improving the responsiveness and reliability of the Orbbec camera system in a ROS2 environment.