Upload data by using AnalyticDB lake storage OSS
When datasets, model weights, or simulation asset files are large and exceed the browser upload limit, you can upload files through Object Storage Service (OSS) and then reference the data by using the OSS path in the Embodied Intelligence Platform. This method is applicable to the following scenarios:
Upload large-scale training datasets (tens of GB or more)
Migrate existing data in bulk from on-premises to the cloud
Share data among team members for collaboration
Prerequisites
Object Storage Service (OSS) is activated.
An AnalyticDB Embodied Intelligence Platform instance is created with
Step 1: Create a RAM user and grant permissions
For security purposes, we recommend that you use the AccessKey of a RAM user instead of that of your Alibaba Cloud account to access OSS.
Log on to the RAM console.
In the left-side navigation pane, choose Identities > Users.
Click Create User.
Enter a User Login Name and Display Name, select OpenAPI Access to generate an AccessKey, and click OK.
Save the generated AccessKey ID and AccessKey Secret.
NoteThe AccessKey Secret is displayed only once when it is created. Keep it safe.
Go back to the user list, click the username of the new user, and on the Permissions tab, click Grant Permission.
Search for and select AliyunOSSFullAccess (manages OSS permissions), and click OK.
Step 2: Upload data to OSS
Select one of the following methods to upload data based on the data size and your preferences.
Method 1: Upload by using ossbrowser (recommended for non-technical users)
ossbrowser is an official graphical management tool for OSS that supports drag-and-drop upload and resumable upload.
Download and install ossbrowser.
Log on with the AccessKey created in Step 1.
Navigate to the target bucket and path, and drag the local files to upload.
Method 2: Upload by using ossutil (recommended for developers)
ossutil is an official command-line tool for OSS that supports batch operations and script automation.
Download and install ossutil.
Configure the AccessKey:
ossutil config -e oss-cn-beijing.aliyuncs.com -i <AccessKey ID> -k <AccessKey Secret>Upload the directory to OSS:
ossutil cp -r /path/to/local/dataset oss://<bucket-name>/datasets/
Method 3: Upload through the OSS console (suitable for small files)
Log on to the OSS console.
In the Buckets, click the target bucket.
Click Objects and then click Upload Object.
Drag files or click to select files, and complete the upload.
Console upload is suitable for single files smaller than 5 GB. For larger files, use ossutil.
Step 3: Reference OSS data in the Embodied Intelligence Platform
After the upload is completed, select OSS as the data source when registering a dataset in the Embodied Intelligence Platform and enter the following information:
Parameter | Description |
OSS Endpoint | The OSS access endpoint, such as |
OSS Bucket | The name of the bucket. |
OSS Path | The path of the dataset in the bucket, such as |
Access Key ID | The AccessKey ID created in Step 1. |
Access Key Secret | The AccessKey Secret created in Step 1. |