ROS与AR.Drone-tum_ardrone-问题和技巧

问题和技巧

问题

• Ar.Drone 1.0版本，压力传感器和磁力计是不使用。
• PTAM初始化失败会导致drone_stateestimation经常崩溃（崩溃发生在PTAM代码）。发生在oarticular，如头两个keyframes之间没有基线。很少发生在praxis。

提示和技巧

1. 相机标定

front, old drone: 0.672049, 0.899033, 0.502065, 0.513876, 0.271972
front, new drone: 0.771557, 1.368560, 0.552779, 0.444056, 1.156010

calibrate with ethzasl_ptam. to work with colored images, in src/CameraCalibrator.cc change:
用ethzasl_ptam做标定，跟有色的图片一起工作，在src/CameraCalibrator.cc改变：

    增加 #include (缺译者注)
    改变 function imageCallback(...) to

void CameraCalibrator::imageCallback(const sensor_msgs::ImageConstPtr & img)
{
    cv_bridge::CvImagePtr cv_ptr = cv_bridge::toCvCopy(img, sensor_msgs::image_encodings::MONO8);

    if(mCurrentImage.size().x != img->width || mCurrentImage.size().y != img->height)
        mCurrentImage.resize(CVD::ImageRef(img->width, img->height));

    memcpy(mCurrentImage.data(),cv_ptr->image.data,img->width * img->height);

    mNewImage = true;
}

参数: c1 to c8

可以评估更容易:

• 记录飞行： rosbag record -O calibFlight.bag /ardrone/image_raw /ardrone/navdata /cmd_vel
• 回放：rosbag play -l calibFlight.bag
• 打开两个状态评估节点：其一重定义名称

rosrun tum_ardrone drone_stateestimation __name:=drone_stateestimationn2 /ardrone/predictedPose:=/ardrone/predictedPose2
rosrun tum_ardrone drone_stateestimation

• 绘制各自的评估值

rxplot /ardrone/predictedPose/dx,/ardrone/predictedPose2/dx

• 使用动态配置

in drone_stateestimation2, use only control gains
in drone_stateestimation, use NO control gains, but instead navdata / speeds / PTAM.

• setting c_i drone_stateestimation2 such that graphs match best (play around).

参数: PID control

approximate in "simulation" based on c1 to c8:

• play back any record, to make stateestimation run (dont play /cmd_vel)

rosbag play -l calibFlightZ.bag --topics /ardrone/image_raw /ardrone/navdata

• run stateestimation, in dynamic reconfigure set only control updates to be used
• run gui and autopilot, and control
• plot control / respective pose e.g. with

rxplot /ardrone/predictedPose/yaw /cmd_vel/angular/z

参考：

• https://github.com/tum-vision/tum_ardrone