For a typical plane or quad with limited battery I can see how the maximum of 700 waypoints would be more than one would ever require.
However, when creating a flight plan in Mission Planner for a Rover (boat), particularly the corridor option, the typical mission has around 3,000 - 4,000 waypoints. Three of these missions are typically carried out in a single day if weather permits.
Breaking each mission into 700 waypoints, effectively takes the ‘autonomy’ out of autonomous as I have just got back from three days of scanning a reservoirs with two craft, and I am exhausted from running backwards and forwards.
This is not a Mission Planner limitation, but the limitation in the controller.
There is simply not enough memory in Pixhawk and Cube Black to store more waypoints. (No waypoints are not stored on the SD card.)
New hardware such as Cube Yellow and Orange will have more memory so there will be a possibility to extend the number of waypoints.
Andras, like I said, shoreline passes in corridor require many thousands of waypoints as lakes and reservoirs have many bays and points creating an irregular path. Air surveys are very simple by comparison.
I’ve created an enhancement request here to cover supporting more waypoints by storing them on the SD card. We discussed this a few weeks ago but I can’t immediately promise when one of us will get to this.
Whenever I hear of a user having to input so many waypoints I wonder if there isn’t an easier way to define what needs doing. We have also discussed adding a “mapping mode” and complex fences and stayout zones combined with Dijkstra’s may also help although I fear that the number of fence points will beyond what we can handle (I’ve only tested up to about 80 fence points).
Agricultural missions such as mowing, weeding, watering, etc., also require many waypoints. While I can imagine modeling these missions with a combination of parameters and fences and exclusion areas (all irregularly shaped), I could also see the variants of these missions outrunning the parameters you might think up. For example, when mowing my lawn with my all-electric mower, I turn the mower blades off using DO_SET_SERVO at the end of one run while doing a u-turn before the next run, then turn them back on when the next run starts. On smaller lawns I am spending 30% of my time in these u-turns and turning the blades off during u-turns helps mow more lawn on a single charge.
I am with you both on this matter. I had raised the issue a couple of years ago in the community and concluded that if I wanted more instructions in the mission, I had to pay for it by upgrading from Pixhawk 1. This was after messing with Pixhawk’s memory allocation instructions without success.
Now I am gathering even Pixhawk 2 / Cube runs a inhibiting limitation.
We need a “limitless” solution and the SD card certainly sounds like the way to go.
In the meantime there is a workaround where you put the uav into guided mode and feed it position coordinate waypoints from a companion computer. You could then in theory have infinite waypoints and if you wanted add some simple logic within the companion computer to adjust the exact waypoints being produced on the fly to avoid an object or return for refuelling when required.
Unattended autonomy is the primary goal of this 4 year project, and if I understand your post correctly, someone would have to be sending each waypoint via Mavlink manually?
No, you would write a script to generate the waypoints pre flight or during flight based on pre defined conditions on the companion computer and feed them automatically to ardupilot such that in guided you are simply telling the craft to move
repeatedly to different coordinates.
If you detected a fault in your aircraft or that you had low fuel for example you would automatically stop feeding waypoints for your mission and start feeding waypoints to return home.
As we have started to put together missions for our Redclaw Aquaculture feeder I have been surprised how quickly the number of waypoints escalates. This was especially so when we were considering using the servo commands to control the feeder mechanism.
We have about 30 ponds, each with 4 sides and we are currently looking at having maybe 5 waypoints per side. Adding in about 50 waypoints for navigating along the transport sections between blocks of ponds gives over 600 waypoints just for navigation alone.
Our route planning will be done programmatically and calculated before each mission as we hope to implement route optimization at some stage. Having a computer calculate the waypoint list makes it easy to generate a large amount of data.
Our thoughts at this early stage is that for larger complex missions that push the limits of the internal memory, it might be more appropriate for this heavy lifting to be done by a companion computer and the waypoints to be fed into the flight controller in guided mode as suggested by Tegwin in post #16. The reasoning here is that with complex missions there may well be other logic and inputs impacting the mission as it is in progress (as also suggested by Tegwin in his post). We are expecting that this will be the case for our application.
Anyone knows if there is any update on this part? Is Ardupilot capable to store waypoints in SD card? I am doing a project with rover that uses cheap flight controller with little internal memories. I need to upload 3000-6000 waypoints.