Below is the VMOSCLOUD Cloud Phone Sensor Dynamic Simulation Guide. This feature allows developers to provide realistic physical environment interactions for cloud phone applications by simulating real sensor data files.

1. Feature Overview

VMOSCLOUD cloud phones support dynamic data simulation for various mobile device sensors. By injecting predefined data files into the system, developers can simulate device motion, location, environmental changes, and biometric features. This is suitable for game testing, navigation simulation, health app development, and more.

2. Supported Sensor Types and Format Details

For easy reference, sensors are divided into four major categories. Use the corresponding format string in the data file.

2.1 Motion & Pose Sensors

2.2 Position & Environment Sensors

2.3 Motion Detection & Biometric Sensors

2.4 Foldable Device Specific Sensors (Foldable)

Only effective when simulating foldable device models (e.g., Galaxy Fold, Mate X).

3. Usage & Configuration Process

Sensor simulation is driven by setting a system property pointing to a local data file.

3.1 Core Commands

Bash

# Set sensor data file path
adb shell setprop persist.sys.cloud.sensor.tpl_dp /data/local/tmp/sensor_data.txt

# Stop simulation (clear property)
adb shell setprop persist.sys.cloud.sensor.tpl_dp ""

3.2 Data File Specification

• File Encoding: UTF-8
• Permissions: File must be readable (recommended chmod 644)
• Size Limit: Maximum 1GB
• Basic Structure: Sensor data lines + delay command lines (delay:milliseconds)

File Content Example:

Plaintext

# Simulate tilting left
acceleration:-5.62:2.94:7.21
gyroscope:-0.62:-0.12:0.12
delay:50

# Simulate tilting right
acceleration:5.62:2.94:7.21
gyroscope:0.62:0.12:-0.12
delay:50

4. Practical Scenarios: How to Simulate

Scenario 1: Simulate Device Rotation (Python Data Generation)

Used to test UI adaptation or game controls during device rotation.

1. Generate Data: Use Python script to create 0–360° rotation data.

   Python

   import math
   
   with open('rotation.txt', 'w') as f:
       for angle in range(0, 361, 5):
           rad = math.radians(angle)
           # Simulate X/Z axis gravity component changes
           acc_x = 9.8 * math.sin(rad)
           acc_z = 9.8 * math.cos(rad)
           f.write(f"acceleration:{acc_x:.6f}:0.000000:{acc_z:.6f}\n")
           f.write("delay:100\n")
   

2. Inject Data:

   Bash

   adb push rotation.txt /data/local/tmp/
   adb shell setprop persist.sys.cloud.sensor.tpl_dp /data/local/tmp/rotation.txt
   

3. Verify: Restart the target app and observe if the interface rotates with the data.

Scenario 2: Simulate “Shake” Gesture

Used to test features like WeChat Shake.

1. Create Data File (shake.txt):

   Plaintext

   # Strong left shake
   acceleration:-19.0:0.0:5.0
   gyroscope:0.0:0.0:8.0
   delay:80
   # Strong right shake
   acceleration:19.0:0.0:5.0
   gyroscope:0.0:0.0:-8.0
   delay:80
   # Return to stationary
   acceleration:0.0:0.0:9.8
   gyroscope:0.0:0.0:0.0
   delay:200
   

2. Execute Simulation:

   Bash

   adb push shake.txt /data/local/tmp/
   adb shell setprop persist.sys.cloud.sensor.tpl_dp /data/local/tmp/shake.txt
   # Launch app for testing
   adb shell am start com.tencent.mm/.ui.LauncherUI
   

5. Best Practices & Notes

Data Optimization Suggestions

• Smooth Transitions: Avoid sudden value jumps. Use interpolation algorithms to generate transition data between states (e.g., stationary → motion) to prevent app logic errors.
• Reasonable Delays:
  • Games/VR: delay:16 (≈60Hz)
  • Navigation: delay:200 – 1000 (1–5Hz)
  • General apps: delay:50 – 100 (10–20Hz)
• Physical Realism:
  • When stationary, accelerometer Z-axis should be approximately 9.8 m/s².
  • Magnetometer strength typically ranges between 25–65 μT.

Common Troubleshooting

Next Steps Suggestion

If you need to capture real complex motion data (e.g., running, driving), it is recommended to write a simple Android tool app, install it on a physical device, use SensorManager to listen to sensors, write the data in the above format to a text file, and then directly import it into the cloud phone. This will provide much more realistic data than script-generated values.