Building a Kotlin Android Robot Control App with NVIDIA Jetson and ROS 2

A practical guide to building an Android operator app in Kotlin that communicates with a robot computer running on NVIDIA Jetson. The Android layer handles human interaction while Jetson and ROS 2 handle robotics workloads.

Architecture

Android / Kotlin
      |
 HTTPS / WebSocket / MQTT
      |
 Robot Gateway
      |
    ROS 2
      |
 NVIDIA Jetson
   |    |    |
Vision Nav Control
      |
 Robot Hardware

1. Create the Android project

Create a Kotlin Android project and add lifecycle-aware coroutines and a networking library appropriate for your gateway.

data class RobotStatus(
    val connected: Boolean,
    val batteryPercent: Int,
    val mode: String
)

2. Model robot commands

Keep the Android API small and explicit.

sealed interface RobotCommand {
    data class SetVelocity(val linear: Double, val angular: Double) : RobotCommand
    data object Stop : RobotCommand
}

3. Create a gateway client

The mobile application should communicate with a controlled gateway rather than exposing ROS 2 internals directly.

interface RobotGateway {
    suspend fun send(command: RobotCommand)
    fun observeStatus(): kotlinx.coroutines.flow.Flow<RobotStatus>
}

4. Expose ROS 2 commands

On the Jetson side, the gateway can translate an approved command into ROS 2 messages such as /cmd_vel. Keep validation and safety checks on the robot side.

Kotlin command
      ↓
Gateway validation
      ↓
ROS 2 /cmd_vel
      ↓
Safety controller
      ↓
Motor controller

5. Build the Kotlin UI

Jetpack Compose can represent connection, battery, robot mode, and control state.

@Composable
fun RobotStatusCard(status: RobotStatus) {
    Column {
        Text(if (status.connected) "Robot Online" else "Disconnected")
        Text("Battery: ${status.batteryPercent}%")
        Text("Mode: ${status.mode}")
    }
}

6. Add an emergency-stop path

An emergency stop should not depend solely on Android. Provide a physical or robot-side safety mechanism and treat the mobile stop button as one additional command path.

7. Test failure conditions

Test:

Network loss
Gateway restart
Stale commands
Low battery
Invalid commands
Robot-side safety timeout
Android process restart

Conclusion

Kotlin is a useful supervisory layer for Physical AI robots: Android provides the human interface while Jetson and ROS 2 perform compute-intensive robotics work. NVIDIA Isaac ROS provides CUDA-accelerated ROS 2 packages for perception and other robotics workloads.