We've been recently playing with multivehicle and Mission Planner, and we thought there is something odd with the implementation.
First of all, in single vehicle, Mission Planner doesn't handle well 2 different mavlink channels for a single vehicle. For example connecting a vehicle by RFD900 in serial, and also by wifi on TCP doesn't work well. Mission planner actually thinks it is two separate vehicles, instead of taking the mavlink vehicle id from the messages and just instantiate one vehicle. In this case Mission planner instantiates a vehicle for each link and everything is messed up.
For multivehicle we have found some strange issues as well. The use case is we have a companion computer running some mavlink router software. So several aircrafts are connected to that ground companion computer, and from that ground companion computer we then forward the traffic to Mission planner, by TCP connection.
This time the vehicles are instantiated correctly, but there is a strange loss in performance when we connect more than 1 vehicle here. The first vehicle is connected properly, the parameters are downloaded super fast, almost instantaneously. But for the next vehicles the link quality drops a lot, and the parameters take forever to download. We checked the messages coming from our mavlink router companion computer and they seemed ok, with correct seq id and all that. Although we wasn't 100% sure it was mission planner fault until we tested another GCS, QGC.
In QGC the above works just perfectly, all the vehicles are connected with same great link quality, and parameters get downloaded just instantaneously, that is why I think there is something odd with mission planner implementation regarding this particular setup of multivehicle.
I would love to be more helpful but the truth is I have no idea at all of Mission Planner, we usually develop over QGC. But anyway I thought it would be interesting to report this issue.
Issue details
We've been recently playing with multivehicle and Mission Planner, and we thought there is something odd with the implementation.
First of all, in single vehicle, Mission Planner doesn't handle well 2 different mavlink channels for a single vehicle. For example connecting a vehicle by RFD900 in serial, and also by wifi on TCP doesn't work well. Mission planner actually thinks it is two separate vehicles, instead of taking the mavlink vehicle id from the messages and just instantiate one vehicle. In this case Mission planner instantiates a vehicle for each link and everything is messed up.
For multivehicle we have found some strange issues as well. The use case is we have a companion computer running some mavlink router software. So several aircrafts are connected to that ground companion computer, and from that ground companion computer we then forward the traffic to Mission planner, by TCP connection.
This time the vehicles are instantiated correctly, but there is a strange loss in performance when we connect more than 1 vehicle here. The first vehicle is connected properly, the parameters are downloaded super fast, almost instantaneously. But for the next vehicles the link quality drops a lot, and the parameters take forever to download. We checked the messages coming from our mavlink router companion computer and they seemed ok, with correct seq id and all that. Although we wasn't 100% sure it was mission planner fault until we tested another GCS, QGC.
In QGC the above works just perfectly, all the vehicles are connected with same great link quality, and parameters get downloaded just instantaneously, that is why I think there is something odd with mission planner implementation regarding this particular setup of multivehicle.
I would love to be more helpful but the truth is I have no idea at all of Mission Planner, we usually develop over QGC. But anyway I thought it would be interesting to report this issue.
Thanks!
Version
1.3.74
Platform
[ x ] All
Hardware type ArduPlane