使用ros的好处就是不用还要研究一下SDK，使用的时候不需要在SDK的example中，挨个那个比较合适的example，然后把那个代码修改一下那过来自己用。这样可把相机接在一个笔记本或者机器人上，跑一圈去采数据，rosbag数据可以以后直接使用。
这里的代码给出使用rviz和rqt完成显示。经过测试发现，DUO 3D需要在使用nvidia显卡的机器上使用。恰好我的PC上没有用N卡，所以我把相机放到机器人上。这边用来显示。PC的版本修改了launch文件，用于在把采集节点放在机器人上，把rqt、rviz等显示放在PC上。
- 双目，可以三位重建，得到点云。
- 内部自带IMU，可以用来完成融合。
- 自带标定工具，这里没有使用棋盘格的形式，使用的是圆形点排列。每次只取40个图像，这样数量是固定的，然后在去挑选合适的位姿点，这样容易失败。可以考虑使用kalibr。
- 需要使用Nvidia卡，如果是在机器人上，有些是没有的，感觉这点比较坑。
- 基线有点短，焦距有点小。
- 如果想要使用ARM版的SDK，还需要加入DUO的developer计划，还得交钱，真是无语。
The duo3d_driver ROS node interfaces with DUO SDK and publishes stereo images, disparity, point cloud, and IMU data from the DUO3D sensor. This package is build on Ubuntu 16.04.1 LTS x64 running ROS Kinetic.
The DUO MLX is an ultra-compact imaging sensor with global shutter and a standard USB interface for ease of use and connectivity. The DUO is intended for use in research, autonomous navigation, robotics and industrial areas. The camera's high speed and small size make it ideal for existing and new use cases for vision based applications.
With a programmable illumination board and built-in IR filters and auto-exposure it allows for precise control of lighting environment, delivering configurable and precise stereo imaging for robotics, inspection, microscopy, visual odometry, human computer interaction and beyond.
The DUO MLX solution consists of:
- Factory Calibrated Stereo Camera
- Industrial Grade Monochrome/Global Shutter Sensors
- Integrated Accelerometer/Gyroscope/Temperature (6 DoF IMU)
- Fully Programmable Active LED Array (3xIR 850nm High Power LEDs)
- DUO SDK License
- DUO Dense3D License
- USB Mini-B Cable
The DUO line of sensors offer a versatile board level solution for 3D sensing, a solution equipped with IMU and IR illumination and a board level sensor with an aluminum enclosure and a complete system for utilizing stereo vision that comes with a DUO MLX sensor and the DUO VPC processor. All DUO sensors are ready to work right out of the box and support a wide range of accessories and configurations.
For more information visit DUO3D Home Page
The DUO SDK package is required for DUO device support under your Linux operating system. It consists of kernel module and user mode libraries that expose all of the DUO functionalities with a simple and concise C/C++ API. Included in the package is the Dense3D™, an SIMD/NEON optimized multi-threaded library that performs the disparity and 3D data extraction from DUO stereo image pair.
- To download the DUO SDK visit the Download page
- To install the DUO SDK follow the Install guide
DUO ROS driver is developed and supported in ROS Kinetic.
Get the package from github and put it in your catkin workspace ''src'' folder:
$ cd ~/catkin_ws/src
$ git clone https://github.com/duo3d/duo3d_driver
$ chmod a+x cfg/Duo3D.cfg
Build the package:
$ cd ~/catkin_ws
$ catkin_make duo3d_driver
$ source ./devel/setup.bash
The duo3d_driver node interfaces with DUO SDK and publishes images, disparity, point cloud, and IMU data from the DUO3D sensor.
- /duo3d_driver/left/image_rect (sensor_msgs/Image) Left camera rectified image
- /duo3d_driver/left/camera_info (sensor_msgs/CameraInfo) Left camera info
- /duo3d_driver/right/image_rect (sensor_msgs/Image) Right camera rectified image
- /duo3d_driver/right/camera_info (sensor_msgs/CameraInfo) Right camera info
- /duo3d_driver/rgb/image_rect (sensor_msgs/Image) Left camera RGB rectified image
- /duo3d_driver/rgb/camera_info (sensor_msgs/CameraInfo) Left camera info
- /duo3d_driver/depth/image_raw (sensor_msgs/Image) Colored disparity image
- /duo3d_driver/depth/camera_info (sensor_msgs/CameraInfo) Disparity info
- /duo3d_driver/point_cloud/image_raw (sensor_msgs/PointCloud2) DUO 3D point cloud data
- /duo3d_driver/imu/data_raw (sensor_msgs/Imu) DUO IMU data
~frame_rate(double, default: 30) DUO image capture frame rate~image_size(vector, default: {640,480}) DUO image frame size~dense3d_license(string) Dense3D license string~gain(double, default: 0%) Image gain value [0, 100]~exposure(double, default: 50%) Image exposure value [0, 100]~auto_exposure(bool, default: False) Image auto exposure value [False, True]~camera_swap(bool, default: False) Image swap [False, True]~horizontal_flip(bool, default: False) Horizontal image flip [False, True]~vertical_flip(bool, default: False) Vertical image flip [False, True]~led(double, default: 10%) LED brightness value [0, 100]~accel_range(int, default: +/-2g (0)) Accelerometer Range [+/-2g (0), +/-4g (1), +/-8g (2), +/-16g (3)]~gyro_range(int, default: 250deg/s) Gyroscope Range [250deg/s (0), 500deg/s (1), 1000deg/s (2), 2000deg/s (3)]~imu_rate(double, default: 100Hz) IMU Data Sampling Rate [25, 500]~processing_mode(int, default: BM (0)) Dense3D Processing Mode [BM (0), SGBM(1)]~image_scale(int, default: ScaleXY (3)) Dense3D Image Scaling Mode [ScaleNone (0), ScaleX (1), ScaleY (2), ScaleXY (3)]~pre_filter_cap(int, default: 28) Dense3D Pre-filter cap [1, 63]~num_disparities(int, default: 4) Dense3D Number of disparities [2, 16]~sad_window_size(int, default: 6) Dense3D SAD Window Size [2, 10]~uniqueness_ratio(int, default: 27) Dense3D Uniqueness Ratio [1, 100]~speckle_window_size(int, default: 52) Dense3D Speckle Window Size [0, 256]~speckle_range(int, default: 14) Dense3D Speckle Range [0, 32]
Make sure that DUO device is plugged in the USB port and it is operating properly.
This example demonstrates the acquisition of rectified stereo image pair from DUO. To launch DUO Camera ROS example, in a terminal run the following command:
$ roslaunch duo3d_driver duo3d_camera.launch
You should see the following:
This example demonstrates the DUO's on-board 6DoF sensor fusion using Madgwick algorithm running at 100Hz. To launch DUO IMU ROS example, in a terminal run the following command:
$ roslaunch duo3d_driver duo3d_imu.launch
You should see the following:
This example demonstrates disparity and point cloud generation using DUO. To launch DUO IMU Point Cloud example, in a terminal run the following command:
$ roslaunch duo3d_driver duo3d_depth.launch
You should see the following:
