/oak_d_ros

Primary LanguageC++

OAK_D Complete Interface for ROS

The OakDInterface class manages the operation of the package

First, the OAK-D is configured using the OakDPipeline class. The desired parameters can be configured in the lib/oakd_pipeline.cpp file (information available at: https://docs.luxonis.com/projects/api/en/latest/references/cpp/)

Images are published using the following classes. Its configuration is done through the oak.launch file:

  • OakDTaskMono class (lib/oakd_task_mono.cpp file): Publish Stereo Images
  • OakDTaskRectified class (lib/oakd_task_rectified.cpp file): Publish Stereo Rectified Images
  • OakDTaskDepth class (lib/oakd_task_depth.cpp file): Publish Depth Images
  • OakDTaskColor class (lib/oakd_task_color.cpp file): Publish RGB/BGR Images

For the use of a network with inference in the color camera:

  1. Set .blob file path and network configuration in oak.launch file
  2. Change in OakDTaskColorNeuralInference class (lib/oakd_task_color_neural_inference.cpp file) the desired label_map

For the use of a network with inference in stereo cameras:

  1. Set .blob file path and network configuration in oak.launch file
  2. Change in OakDTaskStereoNeuralInference class (lib/oakd_task_stereo_neural_inference.cpp file) the desired label_map

To add new features named TASK:

  1. In include/oak_interface/oakd_task.hpp:

    • Add publish_TASK at OakPublishList to enable it in oak.launch
    • Add use_TASK at OakUseList with OAK-D nodes needed
    • Add inx_TASK at OakQueueIndex with OAK-D information needed

  2. In lib/oakd_interface.hpp:

    • Add publish_TASK in OakDInterface::read_param to read from oak.launch
    • Add use_TASK in OakDInterface::create_use_list to use all OAK-D nodes needed
    • Add OakDTaskTASK class in OakDInterface::ownSetUp to add it to tasks_list_

  3. In oakd_pipeline.cpp:

    • Make the desired configuration
    • Configure streams_queue pipeline

  4. Create OakDTaskTASK class that inherits from OakDTask (oakd_task.hpp). It has to override the methods:

    • void start(ros::NodeHandle& nh);
    • void run(std::vector<std::shared_ptrdai::DataOutputQueue>& streams_queue, OakQueueIndex& queue_index, std_msgs::Header header);
    • Where: streams_queue is OAK-D information queue and queue_index are their index in the queue
    • void stop();