光学流动边缘有问题

Hellodaniele.marchetti，感谢您对英特尔Aero Ready to Fly无人机的兴趣。如果您已经在共享的主题上完成了Coline提到的校准，请告诉我们吗？校准工具：http：// wiki。
ros.org/camera_calibrationInstructions:http://wiki.ros.org/camera_calibration/Tutorials/MonocularCalibrationrosrun image_view image_view image：= / u***_cam / image_rawAlso，你能告诉我们你有什么传感器吗？
LeddarOne，Lidar Lite？先谢谢你，Eliza



以下为原文

Hello daniele.marchetti,

Thank you for your interest in the Intel Aero Ready to Fly drone. 

Could you please let us know if you have already done the calibration mentioned by Coline on the thread you have shared? 
Calibration Tool: http://wiki.ros.org/camera_calibration
Instructions: http://wiki.ros.org/camera_calibration/Tutorials/MonocularCalibration 
rosrun image_view image_view image:=/u***_cam/image_raw

Also, could you let us know what sensor do you have? LeddarOne, Lidar Lite? 

Thank you in advance,
Eliza

嗨伊丽莎，
感谢您的支持。
是的，我按照该线程中的说明完成了校准。
结果，图像稍微纠正，但边缘问题仍然存在。
另外，我在无人机上安装了LeddarOne。
我想知道光流算法是否可以正确应用，尽管来自相机的图像在边缘有问题。
谢谢，
丹尼尔



以下为原文

Hi Eliza,
thank you for your support. Yes, I have done the calibration as explained in that thread. As a result, the images are slightly rectified but the issue on the edges is still.
In addition, I have LeddarOne installed on the drone. I would like to know if the optical flow algorithm can be correctly applied although the images from the camera have the issue on the edge.
Thank you,
Daniele

Hellodaniele.marchetti，感谢您提供相关信息！能否请您提供有关如何捕获帧的信息？另外，请您查看底部相机的视频流，看看是否为百分比.https：// github
.com / intel-aero / meta-intel-aero / wiki / 06-Cameras-and-Video＃rtsp-streaming-with-the-camera-streaming-daemon还可以确认相机是否牢牢固定在无人机上？
谢谢你，伊丽莎



以下为原文

Hello daniele.marchetti,

Thank you for the information! 

Could you please provide us the information on how are you capturing the frames? 
Also, could you please check the video stream for the bottom facing camera and see if this is percent. https://github.com/intel-aero/meta-intel-aero/wiki/06-Cameras-and-Video#rtsp-streaming-with-the-camera-streaming-daemon

Can you please also confirm that the camera is firmly attached to the drone? 

Thank you in advance,
Eliza

嗨，
我使用u***_cam节点获取帧，使用节点image_view来显示帧。
结果在右边和下边缘出现问题，如上一个附件中所示。
来自底部相机的视频流不起作用。
只有来自英特尔实感的视频流才有效。
最后，我确认底部相机牢牢固定在无人机上。
PS：
我在计算板和开发地面站上安装了Ubuntu 16.04。
谢谢，
丹尼尔



以下为原文

Hi,
I use the u***_cam node to acquire frames and the node image_view to show the frames. The result has an issue on the right and lower edges as shown in the previous attachment.
The video stream from the bottom camera does not work. Only the video stream from the Intel Real Sense works.
Finally, I confirm that the bottom camera is firmly attached to the drone.
 
PS:
I have Ubuntu 16.04 installed on the compute board and on the development ground station.
Thank you,
Daniele

嗨，
我尝试应用光流算法。我从QGroundControl检查质量在50到200之间，具体取决于地板。
用电机移动无人机解除光学流动似乎是有效的。
为了更好地检查，我应该将光流测量的速度与外部传感器测量进行比较。
然而，目前问题已经出现：使用QGroundControl-widgets-analyze我检查了光流算法（M1：C100：OPTICAL_FLOW_RAD.integrated_x和M1：C100：OPTICAL_FLOW_RAD.integrated_x）中的无人机速度是否在帧中表示
相对于车架旋转180度。
我是唯一一个展示这个问题的人吗？
此外，如果我设置EKF_AID_MASK = 2，如果我用手移动无人机，local_position_x和y也是常数零，因此我无法切换到位置飞行模式。
谢谢，
丹尼尔



以下为原文

Hi,
I tried to apply the optical flow algorithm.ta I checked from QGroundControl stion that the quality is between 50 and 200 depending on the floor. Moving the drone by hand with motor disarmed the optical flow seems to work. For a better check I should compare the velocity measured by the optical flow with an external sensor measurement. However, at the moment an issue has come out: using QGroundControl-widgets-analyze I checked that the drone velocity from the optical flow algorithm (M1:C100:OPTICAL_FLOW_RAD.integrated_x and M1:C100:OPTICAL_FLOW_RAD.integrated_x) is expressed in a frame 180 deg rotated respect to the vehicle frame.
Am I the only one that exhibit this issue?
 
Moreover, if I set EKF_AID_MASK = 2, the local_position_x and y are constant zero also if I move the drone by hand, thus I cannot switch to position flight mode.
 
Thank you,
Daniele

您好Daniele，您使用的是哪种PX4固件？
将EKF2_AID_MASK设置为2应该允许您使用位置控制。
什么是systemctl状态航空光流的输出？问候，耶稣



以下为原文

Hi Daniele,

Which PX4 firmware are you using? Setting EKF2_AID_MASK to 2 should allow you to use position control. What is the output of systemctl status aero-optical-flow?

Regards,
Jesus

嗨，
这是systemctl状态航空光流的输出
●aero-optical-flow.service - 航空光流 
已加载：已加载（/lib/systemd/system/aero-optical-flow.service;已禁用;供应商预设：已启用） 
主动：自2018-09-03 10:22:36 CEST后主动（正在运行）;
13分钟前 
主PID：2254（航空光学） 
CGroup：/system.slice/aero-optical-flow.service 
└─2254/ usr / bin / aero-optical-flow
设置03 10:34:39 aerocomputeboard-CherryTrail aero-optical-flow [2254]：好匹配数：1，所需：20
设置03 10:34:39 aerocomputeboard-CherryTrail aero-optical-flow [2254]：好匹配数：1，所需：20
设置03 10:34:39 aerocomputeboard-CherryTrail aero-optical-flow [2254]：好匹配数：1，所需：20
设置03 10:34:39 aerocomputeboard-CherryTrail aero-optical-flow [2254]：好匹配数：1，所需：20
设置03 10:34:39 aerocomputeboard-CherryTrail aero-optical-flow [2254]：好匹配数：1，所需：20
设置03 10:34:39 aerocomputeboard-CherryTrail aero-optical-flow [2254]：好匹配数：1，所需：20
设置03 10:34:39 aerocomputeboard-CherryTrail aero-optical-flow [2254]：好匹配数：1，所需：20
设置03 10:34:39 aerocomputeboard-CherryTrail aero-optical-flow [2254]：好匹配数：1，所需：20
设置03 10:34:39 aerocomputeboard-CherryTrail aero-optical-flow [2254]：好匹配数：1，所需：20
设置03 10:34:39 aerocomputeboard-CherryTrail aero-optical-flow [2254]：好匹配数：1，所需：20
正如我在QGroundControl-Widgets-analyze中看到的那样，光流是活跃的，质量高于100但LOCAL_POSITION_NED_没有任何输出。
在分析工具中，我看到三个光流领域：
M1：C1：当我向后移动无人机时，OPTICAL_FLOW_RAD.integrated_x为正
M1：C1：当我将无人机向右移动时，OPTICAL_FLOW_RAD.integrated_y为正
M1：C100：OPTICAL_FLOW_RAD.integrated_x和M1：C100：OPTICAL_FLOW_RAD.integrated_y的工作方式类似。
M1：OPTICAL_FLOW_RAD_integrated_x和M1：OPTICAL_FLOW_RAD_integrated_y不起作用
先谢谢你，
丹尼尔



以下为原文

Hi,
This is the output of systemctl status aero-optical-flow
● aero-optical-flow.service - Aero optical flow
   Loaded: loaded (/lib/systemd/system/aero-optical-flow.service; disabled; vendor preset: enabled)
   Active: active (running) since lun 2018-09-03 10:22:36 CEST; 13min ago
Main PID: 2254 (aero-optical-fl)
   CGroup: /system.slice/aero-optical-flow.service
           └─2254 /usr/bin/aero-optical-flow
 
set 03 10:34:39 aerocomputeboard-CherryTrail aero-optical-flow[2254]: Number of good matches: 1, desired: 20
set 03 10:34:39 aerocomputeboard-CherryTrail aero-optical-flow[2254]: Number of good matches: 1, desired: 20
set 03 10:34:39 aerocomputeboard-CherryTrail aero-optical-flow[2254]: Number of good matches: 1, desired: 20
set 03 10:34:39 aerocomputeboard-CherryTrail aero-optical-flow[2254]: Number of good matches: 1, desired: 20
set 03 10:34:39 aerocomputeboard-CherryTrail aero-optical-flow[2254]: Number of good matches: 1, desired: 20
set 03 10:34:39 aerocomputeboard-CherryTrail aero-optical-flow[2254]: Number of good matches: 1, desired: 20
set 03 10:34:39 aerocomputeboard-CherryTrail aero-optical-flow[2254]: Number of good matches: 1, desired: 20
set 03 10:34:39 aerocomputeboard-CherryTrail aero-optical-flow[2254]: Number of good matches: 1, desired: 20
set 03 10:34:39 aerocomputeboard-CherryTrail aero-optical-flow[2254]: Number of good matches: 1, desired: 20
set 03 10:34:39 aerocomputeboard-CherryTrail aero-optical-flow[2254]: Number of good matches: 1, desired: 20
 
Optical flow is active as I can see in QGroundControl-Widgets-analyze and the quality is above 100 but LOCAL_POSITION_NED_ doesn't exhibit any output. In analyze tool I see three fields of optical flow:
 
M1:C1:OPTICAL_FLOW_RAD.integrated_x is positive when I move the drone backward
M1:C1:OPTICAL_FLOW_RAD.integrated_y is positive when I move the drone the right
 
M1:C100:OPTICAL_FLOW_RAD.integrated_x, and M1:C100:OPTICAL_FLOW_RAD.integrated_y work similarly.
 
M1:OPTICAL_FLOW_RAD_integrated_x and M1:OPTICAL_FLOW_RAD_integrated_y don't work
 
Thank you in advance,
Daniele

嗨Daniele，您可以试试附加的px4固件吗？
请注意，不要将SENSE_EN_LEDDAR1设置为1，而是需要将AEROFC_TELE_BPS设置为0.确保执行帧复位并校准所有组件。
同时将EKF2_AID_MASK设置为2和EKF2_HGT_MODE = 2（范围传感器）.Regards，Jesus     
aerofc-v1_default_leddar_one.px4.zip 
561.0 K.



以下为原文

Hi Daniele,

Could you try with the px4 firmware attached. Please note that instead of setting SENSE_EN_LEDDAR1 to 1, you need to set AEROFC_TELE_BPS to 0. Make sure to do a frame reset and calibrate all the components. Also set EKF2_AID_MASK to 2 and EKF2_HGT_MODE = 2 (Range Sensor).

Regards,
Jesus

•   aerofc-v1_default_leddar_one.px4.zip 561.0 K  
  
  

嗨耶稣，
感谢您的支持。
我上传了无人机，附带了你的px4固件。
当我从QGroundControl连接到无人机时，它显示以下消息：
您当前已连接到车辆，但未返回完整参数列表。
因此，全套车辆设置选项无法使用。
因此，我无法校准我的车辆并重置机身。
但是我可以设置AEROFC_TELE_BPS = 0，EKF2_AID_MASK = 2和EKF2_HGT_MODE = 2.通过这种设置，光流工作，我从LOCAL_POSITION_X和Y得到一致的值，所以它可能是正确的方法。
谢谢，
丹尼尔



以下为原文

Hi Jesus,
thank you for your support. I uploaded the drone with your px4 firmware attached. When I connect to the drone from QGroundControl, it shows the following message:
You are currently connected to a vehicle but it did not return the full parameter list. As a result, the full set of vehicle setup option are not avaliable.
Thus I cannot calibrate my vehicle and reset the airframe.
However I can set AEROFC_TELE_BPS = 0, EKF2_AID_MASK = 2 and EKF2_HGT_MODE = 2. With this setup, the optical flow works and I get consistent values from LOCAL_POSITION_X and Y, so it could be the correct way.
Thank you,
Daniele

嗨，
有兴趣，我已经阅读了这篇文章，因为我们遇到了同样的问题。
我们正在尝试使用面朝下的相机和LeddarOne来实现光流。
到目前为止没有结果，因为LOCAL_POSITION_NED.x和.y保持0。
然而，在闪烁了aerofc-v1_default_leddar_one.px4后，它正在工作。
唯一的事情是VISION_POSITION_ESTIMATE的融合已停止在此版本中工作，该版本在AeroFC固件版本1.8.0版本中有效。
是否有计划将此图像中的更改/修复提交给github中的源，因此我们可以自己构建它并使用光流运行？
感谢致敬，
Gertjan
Florinco BV



以下为原文

Hi,
 
With interest I have read this post, as we are experiencing the same problem.
We are trying to get optical flow working, using the down facing camera and LeddarOne. So far no result, because the LOCAL_POSITION_NED.x and .y stay 0.
 
However after flashing aerofc-v1_default_leddar_one.px4 it's working. The only thing is that fusion of the VISION_POSITION_ESTIMATE has stopped working in this build, which was working in release AeroFC firmware version 1.8.0.
 
Is there any plan on commiting the change/fix in this image to the sources in github, so we can build it ourself and run with optical flow?
 
Thanks and regards,
 
Gertjan
Florinco BV

由于我们的项目依赖于光流和视觉姿势估计（来自我们自己的室内GPS），我们对此问题的任何更新都非常好奇。



以下为原文

Since our project depends on optical flow in combination with vision pose estimate (coming from our own indoor GPS), we are very curious about any updates on the issue.

最后我决定购买PX4flow。
我将用它实现光流，因为它是推荐的方法。



以下为原文

Finally I decided to buy PX4flow. I will implement optical flow with it because it is the recommended method.

我提供的飞行控制器固件（aerofc-v1_default_leddar_one.px4）的源代码可在此处找到：https：//github.com/zehortigoza/PX4Firmware/tree/stable-leddar_one我建议在PX4 Github问题部分打开票证
用最新的PX4版本调查问题。回答，耶稣



以下为原文

The source code to the flight controller firmware I provided (aerofc-v1_default_leddar_one.px4) can be found here:
https://github.com/zehortigoza/PX4Firmware/tree/stable-leddar_one

I would recommend opening a ticket on the PX4 Github Issues section to investigate the issue with the latest PX4 release.

Regards,
Jesus