Video

Solved! The problem was solved by installing the latest versions of the packages just in case. I am posting a successful version just in case. Unity was version 2022.3.32, AR Foundation was 5.1.4, AR Core XR Plugin was 5.1.4, cesuim was 1.10.1, and ar core extension was 1.44.0. The problem that occurred before does not occur and we will try to address it further. thank you!!

Excuse me, but I have a few questions. I tracked the log, and the camera keeps showing a log saying 'waiting for ar session to become stable'. Is this normal? (It's not an error message, it's just a log.) Also, in Korea, where I am, Geospatial is not available in 3D. It's not recorded, but I'd like to know if Geospatial can't be used in such a space. Sorry to bother you, but I've been trying to resolve this for several days now. I would appreciate your help.

I will put a sample demo together for this and reply here, if I don’t feel free to email me but I will try to get to it tomorrow.

@@dilmerv any news for this?

You can download it from the video description through GitHub.

Take a look at these two resources: - Video about imitation learning which should drastically help with faster training: czcams.com/video/kpb8ZkMBFYs/video.htmlsi=Foe8FnbPabfj2K9A And also the docs here: github.com/Unity-Technologies/ml-agents/blob/develop/docs/ML-Agents-Overview.md#imitation-learning Thank you for your comment!

Take a look at these two resources: - Video about imitation learning which should drastically help with faster training: czcams.com/video/kpb8ZkMBFYs/video.htmlsi=Foe8FnbPabfj2K9A And also the docs here: github.com/Unity-Technologies/ml-agents/blob/develop/docs/ML-Agents-Overview.md#imitation-learning Thank you for your comment!

@@dilmerv Thank you very much for your reply, but I have found that the improvement learning in mlagents is just a step-by-step teaching of machine learning by humans, without any feedback, that is, it cannot "correct" the behavior of the machine, but only input the desired action step by step to the machine. Can mlagents achieve "real-time correction" of the machine's behavior? That is to say, there is a process of feedback.

Take a look at these two resources: - Video about imitation learning which should drastically help with faster training: czcams.com/video/kpb8ZkMBFYs/video.htmlsi=Foe8FnbPabfj2K9A And also the docs here: github.com/Unity-Technologies/ml-agents/blob/develop/docs/ML-Agents-Overview.md#imitation-learning Thank you for your comment!

@@dilmerv Thank you very much for your reply, but I have found that the improvement learning in mlagents is just a step-by-step teaching of machine learning by humans, without any feedback, that is, it cannot "correct" the behavior of the machine, but only input the desired action step by step to the machine. Can mlagents achieve "real-time correction" of the machine's behavior? That is to say, there is a process of feedback.

That's a great idea, let me give this some thought but thank you for your feedback. As far as making it work, are you using MRTK 3? or MRTK 2?

@@dilmerv Hi Dilmer thankyou for your reply , I am using MRTK3 , with unity version 2022.3.3, with using MixedReality Toolkit in the project . Want to convert it to Meta Quest 2 and 3..

Hey thanks for your question. Can you check your URP profile and make sure HDR is uncheck? Also did you use the Magic Leap Setup tool to configure your project?

If you use PolySpatial then there is an option to use Play To Device which works for the simulator and also a remote app on the device. Without PolySpatial which is what I did here, I tested it mostly by deploying to the actual device. Great question!

Hey thanks for your message. Could you verify that you have the following or greater versions? - Xcode version 13.0 or later - Cocoapods 1.4.0 or later if using Cocoapods

@@dilmerv Thanks for the reply! I'm using Xcode 15.4 and Cocoapods 1.12.1

Great question. MARS uses ARKit behind the scenes which means only the rear-camera is supported for body tracking. I recommend looking into something like mediapipe github.com/google-ai-edge/mediapipe/blob/master/docs/solutions/pose.md

To be honest I am not sure, normally I use the same account across devices so I wouldn't be able to tell you. But as a dev yes that would be amazing if you do so since most times we don't use the same account during development for other companies.

Hey that’s a smart move, thanks for staying up to date and for your feedback Lair !

Hey this is a great question. This is what Magic Leap responded to your question which makes sense to me as how it worked when I tested it. From Magic Leap “After scanning the space with spaces, the map will be saved to the device. Once the device is a localized into the space. It will try to re-localize to the space the next time it is reboot. The device tries to gain localization in the background , if the space was unchanged, the device will be able to localize i to it very quickly.” Let me know if you have other questions, thanks for your time.

VRChat has contributed a ton to it, completely agree!

Great question and thanks for your comment. Currently, the Spatial Anchors will work with the devices mentioned through AR Foundation support, but the persistent aspect won’t unless you add other services such as Google Cloud Anchors or similar that will handle the native localization and storage part.

@@dilmerv many thanks! 🙏

Hey how are you! Mmm what part of it is not working? Are you getting any errors? Let me know and I can try to test it this weekend.

@@dilmerv Unity 2022.3.32 + Meta XR Interaction SDK OVR Samples v.65 + Scene HandGrabPoseTool + Meta Quest 3 (dev mode) connected with OculusLink. Unity Editor in Play mode cannot "record" grab poses, or better, no Hands are rocognized at all.

Hey, thanks for letting me know. I did try to keep a balance between regular-paced coding and fast-forwarding, and I'm glad you noticed.

Hey great question, and yes Meta provides Spatial Anchors as well and they are also persistent. You can find more info about them here: developer.oculus.com/documentation/unity/unity-spatial-anchors-persist-content/ Thanks and let me know if you've further questions.

@@dilmerv thanks for the info man! do you plan to play this topic on your channel and test it as well?

I am glad you found this video useful, thank you for letting me know. If you ever seek specific XR content let me know!

@@dilmerv thank you, I will take this into account)

Hey Juan thanks for your comment and for watching this video. Take a look at this video which should be the first one of this series: czcams.com/video/KqH0zv3e2AY/video.html Let me know if you've additional questions.

Hey thanks again for your feedback Lair !