Import data manually into Azure Spatial Anchors to display a virtual object

Kastler 36

I am looking to display a virtual object in the real world using Azure Spatial Anchors technology. To do this, I have this virtual object and a point cloud of the scene made with a laser scanner. I would like to know if it is possible to manually integrate these data into the Azure Cloud so that the virtual object can be displayed in the scene later. To do this, I would use Unity to define an anchor (center of the coordinate system) relative to the virtual object and relative to the point cloud. These data would then be imported into the Azure Cloud, and I could use it to have Azure Spatial Anchors match this point cloud with the feature points found in the scene during the session. Does Azure Spatial Anchors allow this?

António Sérgio Azevedo 7,661 Reputation points Microsoft Employee

2022-01-11T16:54:06.977+00:00

Same question in StackOverflow: https://stackoverflow.com/questions/70665451/import-data-manually-into-azure-spatial-anchors-to-display-a-virtual-object
António Sérgio Azevedo 7,661 Reputation points Microsoft Employee

2022-01-12T11:04:14.607+00:00

@Kastler From my initial research I believe what you are trying to achieve is possible, using the right mix of Locate Strategy (eg. Using NearAnchor) and then leverage UpdateAnchorProperties. I have contacted Product Team to confirm.
Kastler 36 Reputation points

2022-01-12T19:00:03.837+00:00

Thank you for your answer. To be clear, what I really want to know is if it is possible to manually import this point cloud into the Azure cloud, in order to reuse it to compare this point cloud and the characteristic point cloud detected during a session. From what I understand, Azure Cloud encrypts the data, and I don't know if that would be a problem.
António Sérgio Azevedo 7,661 Reputation points Microsoft Employee

2022-01-14T14:13:37.033+00:00

Hello @Kastler can you be more specific on what you mean by "manually import this point cloud into the Azure cloud" ? Would you like to use for instance Azure Blob Storage?

Thanks!

1 answer

António Sérgio Azevedo 7,661 Reputation points Microsoft Employee

2022-01-17T11:17:40.877+00:00
Hello @Kastler ,
I have received confirmation from Product Team that unfortunately ASA does not support users to upload their own point clouds for matching later. Instead ASA collects the environment information from the native platform tracker at the time of creating spatial anchor and use it later for matching.

We will consider this scenario in future iterations if demanded by our customers, for that can I kindly ask you add a new Feature Request in feedback.azure.com | Azure Spatial Anchors and share it here so others can vote for it as well?

Appreciate your time!

Remember:

Please accept an answer if correct. Original posters help the community find answers faster by identifying the correct answer. Here is how.

Want a reminder to come back and check responses? Here is how to subscribe to a notification.
Please sign in to rate this answer.

2 people found this answer helpful.
Kastler 36 Reputation points

2022-01-20T19:07:15.473+00:00

Hello @António Sérgio Azevedo ,
I must admit that I am surprised by your answer. Spatial anchors are stored in the Azure cloud, and the Azure cloud allows the user to import and export data. Could you explain why this can't be done here ? I know that the data in the Azure cloud is encrypted, is it related to that?

António Sérgio Azevedo 7,661 Reputation points Microsoft Employee

2022-01-21T14:56:01.077+00:00

Hello @Kastler ,

I think we need to separate two concepts in your question, let me describe those on my own words so that you can confirm I understood them well:

1) In the original question, you asked if it was possible to upload point clouds of a "virtual" scene that would be displayed when a user interacts with his\her device on the "real" scene. And the answer is no, you can't upload your own point clouds - ASA collects the environment information from the native platform tracker at the time of creating spatial anchor - this is, you need to be there physically in the "real" scene.

2) The other question that arose was if Azure Spatial Anchors encrypts point clouds when storing them and if that is the reason why you can't upload your own point clouds - and the answer is no. The encryption\decryption of the data points is part of ASA internals and not relevant in this situation. In the end of the day you are dealing with a REST API abstracted with the use of Azure Spatial Anchors SDK.

I hope I could help. Let me know.

Kastler 36 Reputation points

2022-01-21T17:54:31.357+00:00

Hello @António Sérgio Azevedo ,
Ah, maybe we have misunderstood each other. In the original question, I am talking about a point cloud that I have acquired with a laser scanner, directly on the field (real scene). I am not trying to display this point cloud. What I want to display is a virtual object that I would have attached to this point cloud using Unity.
To display this virtual object in the real world with ASA, I need to import manually my point cloud into Azure Cloud, so that ASA can use it and map it to the feature points detected in real time in the scene.
So I'm trying to find out if this is possible.

António Sérgio Azevedo 7,661 Reputation points Microsoft Employee

2022-01-24T10:57:56.703+00:00

Thanks for clarifying the question, I am much more clear now :)!

So my answer still applies: "ASA collects the environment information from the native platform tracker at the time of creating spatial anchor and use it later for matching." Unfortunately this is not yet a feature in ASA (to upload your own custom cloud points captured without using the native platform tracker). And this is not related with data being encrypted on Azure.

As I mentioned before, I can reassure that we are looking at all feedback coming in feedback.azure.com | Azure Spatial Anchors . Can you add this feature idea there and share here as well? We will consider it for next iterations of the service if others vote for it as well.

Let me know if you have further questions?
All Best!

Kyle 0 Reputation points

2023-04-06T05:19:05.99+00:00

A little late to the party, but I'm just getting into Azure Spatial Anchors on the HoloLens and learning how different devices can anchor an object to the same location in physical space. A couple of questions:

Only point cloud data is used when the HoloLens is determining if an anchor should be placed in a specific physical location, correct? No GPS is used to determine where the user is located in the world? If this is the case, and I had one room in California and an exact replica in New York, and I upload an anchor placed on a couch in the California room, someone who downloaded the anchor in the New York replica room should have it attached to the same physical object, correct?

I've been scouring the Internet for info on sharing anchors over a local network but haven't had much luck. I'm using Unity 2020.3.25f1 LTS and read about using the WorldAnchor component and a WorldAnchorTransferBatch to share the data. I didn't have any luck on that Unity version - I didn't have access to the WorldAnchor component even after importing all of the required packages. (I later tried it on 2019.4.4f1 and had access to the WorldAnchor component, but I need v2020 for my project). I even tried using the WorldLockingTools package and sending the frozenWorldState.hkfw file over the network to another HoloLens to load, but that unfortunately didn't work. That being said, is it possible to send the point cloud data used by Azure Spatial Anchors over a local network and have another device download the data from there instead of relying on the Azure cloud? Is it even possible to view the raw point cloud data on the Azure Cloud?

I really appreciate you taking the time to read this. Thank you for your help!

Rob Caplan - MSFT 5,412 Reputation points Microsoft Employee

2023-04-06T06:04:47.93+00:00

@Kyle, please post new questions in their own thread, one question per thread, rather than adding them onto existing threads on different subjects as comments.
Sign in to comment