DJI Agras T25 for spotspraying: trouble calculatiing right amount of herbizide needed

Hi,
We are planning to conduct herbicide field trials with the DJI Agras T25 drone. We want to test whether the drone could also be used for spot or patch spraying. Although weed mapping works reasonably well with the Magic Tool, it is time-consuming. We’re also having trouble calculating the amount of herbicide to apply. Are the quantities calculated by Pix4D based purely on area? As soon as the mission is loaded onto the DJI remote control, the required herbicide amount increases significantly. The actual amount applied is then significantly reduced, but it’s still considerably more than Pix4D’s prediction.
We tested several variations: the mapping was always done in a 0.25-metre grid and the application map had spots buffered once at 1 metre and once at 4 metres (our assumed drone working width). Using a smaller spot size and narrower spacing for the flight plan resulted in the least amount of herbicide, which was still much more than Pix4D suggested.

My question at this point is: is this a DJI issue and do I just need to figure out which row spacing and spot size yield matching quantities? Or are there already settings in Pix4D that help with the actual calculations?

Best Regards

Enrico

Hi Enrico,

That’s just the reality of map-based prescriptions. And this type of behavior is more or less true for all types of machinery. We as a map provider / you as the map creator, cannot know the physical limitations of the applying machinery, meaning the machine in itself has delays, internal buffers, and resampling of prescription maps we supplied, which will lead to different final amounts applied.

You can mitigate this problem to some degree if you plan accordingly based on the capabilities of the machinery, adjusting the grid size and grid direction to the actual application direction and nozzle configuration of your machine.

Even though DJI works with a 1 m by 1 m grid, the swath of the drone is not 1 m by 1 m. It’s much bigger, so there will be oversampling → Resulting in more product needed than what is displayed in Pix4Dfields.

What you can try is to plan with a realistic grid size according to the swath width of your spreader.That should give you much better amount predictions.