microsoft / AirSim

Open source simulator for autonomous vehicles built on Unreal Engine / Unity, from Microsoft AI & Research
https://microsoft.github.io/AirSim/
Other
16.44k stars 4.58k forks source link

HI Team, #3389

Open chiragkhanna28 opened 3 years ago

chiragkhanna28 commented 3 years ago

Question

What's your question?

Include context on what you are trying to achieve

Context details

Include details of what you already did to find answers

chiragkhanna28 commented 3 years ago

Hi Team,

We are implementing Microsoft Airsim for one of our client for Insurance project. We would like to simulate the farm images using drone and derive the crop and soil quality of it by sending the images captured to custom vision.

Is it possible to achieve above use case using AirSim. If yes can you please help with the templates which i should use in Epic Games launcher, which can give me nice images for the farm. Can we also do geofencing on Airsim.

And Is it possible to have different images which depicts bad areas to cultivate crop.

Looking forward to hear from you.

Regards, Chirag Khanna

jonyMarino commented 3 years ago

Hi, @chiragkhanna28, and welcome! Yes, you can simulate all that with AirSim, you will need to manage to send the images to a server. Unfortunately, we don't have environment templates to provide, you will have to search on your own. With regard to "And Is it possible to have different images which depicts bad areas to cultivate crop." yes, of course, you can model that inside unreal.

chiragkhanna28 commented 3 years ago

Hi @jonyMarino ,

Thanks for your response. We have created the custom project on which we have simulated farming environment as per our requirement.

We are struggling on how to move the drone over the rectangular area farm, so that we can capture the completed area over which drone is flying. We tried to use the hello_drone.py file inside Airsim master repository to simulate flying of drone using python script, but could not find the method from the core api which can be used to fly the drone over the complete rectangular area. Can you please help us on the same and we would also like to capture images of ground view in block size of 10x10 such that area occupied by one image of 10x10 should not overlap with the previous block of 10x10. Can we achieve this.

We would also like to get images while drone is running over the field in farm after regular intervals and send these images to Custom vision by running it as a container on simulator. Is it possible to run IOT edge on the simulator (please note inside IOT edge we have run custom vision service as a container successfully) and if yes, can you also please give some insights on how we can achieve this.

Thanks in advance, Chirag Khanna

jonyMarino commented 3 years ago

Hi @chiragkhanna28! With this script, you can fly (with an Xbox controller, for example) and collect the points you are interested in: https://gist.github.com/zimmy87/fd2ff063823d532a68388ecee803ff69. Or take the from the Unreal editor and transform them to NED coordinates.

Then you can use moveToPositionAsync or moveOnPath to follow the collected points in a path. And you can take images in the middle with another script or thread or between calls to moveToPositionAsync. I will ask about IOT edge.

chiragkhanna28 commented 3 years ago

Hi @jonyMarino,

Thanks for your response. We are trying to split the area over which drone is flying into smaller pieces. For that we need that when our drone is running in the simulator mode, then in the first time it will capture image of block size let say 10x10, then the second time it again capture image of block size 10*10 but this time the captured image should not get overlap with the first image, so that we have 2 mutually exclusive and continuous section of field. Similarly we keep on repeating this process and we can get bunch of images which gives the complete area divided in smaller sections.

Please let us know how can we achieve this.

Thanks in advance, Chirag Khanna

jonyMarino commented 3 years ago

Hi @chiragkhanna28! were you able to progress with this?