Feature/time sync #3
Conversation
The delay is usually 1-2 ms but can be upto 20 ms
| if(bc_data.timeStamp.nanoTime < ::g_prevNanoTime){ | ||
| ::g_counter++; | ||
| } | ||
| g_hardwareStamp = ::g_counter*::g_WRAP_TIME + double(bc_data.timeStamp.nanoTime)*(1e-9); |
There was a problem hiding this comment.
The cuckoo_time_translator packages provides a class that does this for you. It is also used within the update call up in line 64. If the nanoTime proves to be a proper wrapping hardware clock you can just use it instead of bc_data.timeStamp.time in line 64 - of course you would need to update the information given to the constructor of the device_time_translator.
| dji_sdk::A3GPS A3_GPS; | ||
| dji_sdk::A3RTK A3_RTK; | ||
| // Declare Time translator | ||
| boost::shared_ptr<cuckoo_time_translator::DeviceTimeUnwrapperAndTranslator> device_time_translator_; // TODO: Why does this have to be a boost::shared_ptr ? |
There was a problem hiding this comment.
The boost::shared_ptr is not needed. In the example driver this was the simplest available solution (pre c++11) to the deferred initialization requirement that came with the fact that the node handle was it requires was created in the onInit method.
As your class gets the node handle from outside passed to the constructor you don't need a deferred initialization at all and could make this a simple member and move its initialization to the initialization list of the constructor.
| @@ -50,10 +52,19 @@ void DJISDKNode::broadcast_callback() | |||
| auto current_time = ros::Time::now(); | |||
|
|
|||
|
|
|||
|
|
|||
There was a problem hiding this comment.
Did you check that bc_data doesn't have a receive time stamp? If it was high quality it clearly would..
If not consider moving up the ros::Time::now() call right below the getBroadcastData() call.
|
@HannesSommer thanks for the review, I'll incorporate the changes before merging to master :) |
|
@raghavkhanna , wait with merging for this PR to settle: ethz-asl/cuckoo_time_translator#5. There are also some new changes (last commit) that I introduced to go better in line with the new drivers I saw. As a blueprint for the necessary adaptations : ethz-asl/pointgrey_camera_driver@5dac172. |
|
Superseded by #8. Closing. |
Added hardware time stamping and RC based roll/pitch/yawrate/thrust control
@inkyusa @marija-p