The control home is the screen you drive the robot from. Open it with Control on the Hub. The whole screen is the viewport — a camera feed or the 3D view — with a motion card on the left, an OBSERVE card on the right, and two virtual joysticks in the lower corners when no gamepad is connected.
DANGER
Every motion button on this screen moves the real robot. Keep people, animals, and obstacles at least 2 m clear of the robot whenever it is operating. Check your surroundings before you press anything, and keep a hand near E-STOP.
The top bar floats over the video in two groups. From left to right:
| Element | What it shows / does |
|---|---|
| Logo | Opens the screen menu — Hub, Control, Dashboard, Settings, Maintenance, Logs, Media, Guide, Update. Leaving the control screen stops the robot where it is; it stays connected |
| Status light | A green, amber, or red light that pulses slowly — fast when something is critical, and a steady red while not connected. Tap it for Overall system status (see Connection) |
| Gait chip | The robot's current gait name, such as STANDING, colored by state. Hidden while not connected |
SLAM chip | Appears only while external control (SLAM or SDK) is active. Tap it to hand control back to the app |
| Signal bars | WiFi signal strength, and how fast data is moving over the robot link |
| Log badge | Unread log alerts |
| Robot battery, device battery | The robot pack's charge, with a charging mark while it charges, and your own device's battery. Tap the robot battery for Battery details — voltage, time to 80% on CAN-FD robots, and charging warnings |
E-STOP | Emergency stop — see below |
| Quit | Desktop build only. Asks Quit the app? first, then closes the app. The robot keeps running |
If the robot drops off and does not come back within 8 seconds, the app takes you to the Hub on its own, because that is where you reconnect. It also goes back to the Hub the moment the connected robot changes, so you never end up driving a robot you did not pick.
Going back (the Android back button, or Esc on a PC or Steam Deck) takes you from this screen to the Hub, the same as Hub in the logo menu. The gamepad B button is a robot button on this screen, so it does not leave it. On a PC or Steam Deck, B goes back on screens that don't use it as a robot button. Going back from the Hub asks Quit the app? first (Android and desktop).
E-STOP never fires on a single tap. Pressing it opens the confirmation Emergency stop?, which warns that The robot immediately stops all motion and cuts power. Reinitialization is required to resume control. Press Stop to go ahead or Cancel to back out. The button behaves the same way on every screen, even with a menu open.
WARNING
After an emergency stop the robot has no motor power. Bring it back with Auto Start (below) before it can move again. The step chips show you where the sequence stopped.
Only one device controls the robot at a time. When another device holds control, the status light turns amber and the Ownership row in Overall system status reads Take ▸ — tap it to take control back. While a SLAM or SDK program is driving, you get the SLAM chip instead.
While another device holds control, this device spectates: the screen stays visible, the controls below the top bar are locked, and Spectating at the top offers to take control. See Spectating.
Until the robot's motor control is running, a green Auto Start pill sits at the bottom center of the viewport, grayed out while not connected. Tap it to open the Auto Start dialog. The dialog always stays centered on the screen: when the screen is short, only the Startup sequence list inside it scrolls and the buttons stay visible. On a short screen such as a phone in landscape, each step shows as one line — tap it to open or close its details. The running step and any failed or warning step open by themselves.
You can also start the same sequence from the status light: Overall system status › ▶ Auto start (full sequence).
The dropdown at the top center of the viewport picks what you see. 3D, a live model of the robot, is always there. Camera sources show up only when the robot reports having that camera: Front, Rear, Left, Right, Stairs, Quad, Panorama, CCTV, Thermal, PTZ Mix, and SLAM on LiDAR robots. Tap outside the list to close it without changing the source.
On CAN-FD robots, when the bottom status LEDs are on, the 3D view in 3D and Stairs shows the same colored light. It lights the LED spots under the body, the legs and the floor at foot level, and a blinking LED blinks at the same pace. Each LED casts its own circle on the floor; the two circles overlap slightly off-center, and where they overlap the colors mix. The floor light shrinks as the body lowers. It is not drawn while LiDAR or heightmap points or the obstacle map are shown, so they do not overlap.
Pick a PTZ-type source and the pan/tilt/zoom panel appears over the video. Drag it by its handle to move it.
While docking runs, the selector reads Dock and will not open — the viewport belongs to Dock View until you close it.
Everything that makes the robot move lives on the left.
SIT, STAND, WALK, STAIRS, and DOCKING (only while docking is available), plus Motion, the dashed button that opens the full motion grid. Whichever button matches the robot's current gait is highlighted. See Motion commands.▾ at the bottom tells you it scrolls.Arm operation controls are shown only when the arm 48V rail is on and all six arm joints communicate. Turning arm power off hides these controls. Manual inputs are canceled on power loss. You can leave the manual workspace without requesting Folding; after power returns, start manual control again explicitly. Simulation has no power distribution unit, so the power condition is not applied there.
The Arm menu holds two screens, Manual control and Arm motions, and four basic poses: Folding, Ready, Straight and Lift. The poses move the arm from the menu itself without opening a screen, and are available only while arm control is started and the arm position lock is released.
Choose Arm → Arm motions to open the motion list and the live robot-and-arm 3D view. The list contains Greeting, Multi joint, Auto Recovery and Damping Mode.
The robot rejects a motion when the first one is not the ready pose. Until you press Ready pose once after entering the screen, the other motions appear disabled and a notice is shown. Leaving and re-entering the screen requires pressing it again. Demo motions swing the arm widely, so clear the area before running them.
Choose Arm → Manual control to open the body-front camera beside one combined robot-and-arm 3D control view. The gripper is highlighted in amber. EE focus follows the measured arm tip at a close distance, with a live full-robot 3D PIP in the upper right. Tap that PIP or Full view to see the whole robot; EE focus returns to the close view. Both views use the same measured robot model. EE coordinate axes use the selected control frame and scale with the view size and zoom. Arrows and touch targets are wider, and the default length is 140% of the previous size. Open Input feedback in the 3D header and use its Axis size slider or Smaller axes / Larger axes to choose 100–200%; Defaults restores 140%. This device remembers the setting. Axis size is limited near view edges to prevent clipping. Opening the setting stops current input; changing the display size does not change jog speed or the control frame. The same setting lets you compare A (axis glow), B (a tether from the grab point to your finger), C (a direction ribbon), and B+C (tether and ribbon). This device remembers the choice. Grabbing an axis shows it as held; dragging beyond the small screen dead zone shows the +/− input direction. Rotation uses a curved indicator, or the existing up/down gesture direction when the axis cannot be resolved in the current view. Releasing or canceling input removes the indicator. These graphics show user input, not confirmed robot motion, physical stopping, travel distance, speed or force. The hand camera remains in its own bottom-right PIP within the front video; both camera images use one composed stream.
On narrow screens the camera sits above the combined 3D view. Controls stay in a fixed bottom bar, with larger −/+ and gripper touch targets. Button controls and Axis drag do not resize or remount the views. On web and desktop, holding a jog button keeps the input pressed even if the finger slides outside its bounds; lifting, touch cancellation, lost pointer capture, a second touch, focus loss or disabling the control cancels the hold. Space/Enter can hold a focused button; releasing the key stops it. Routine captions are omitted, while connection, power, error and missing-pose/frame states remain visible. Enlarge 3D swaps camera and 3D panel sizes; Enlarge camera restores them. On wide screens the divider resizes those columns and supports the left/right arrow keys. Changing views or resizing cancels held input.
Opening this screen does not move the arm. Press Start manual control to request manual control, which may move the arm to its ready pose. The EE and gripper controls become available when the robot reports manual control. Grab an arrow at the highlighted gripper in the combined 3D view and drag along its axis. In Rotate, grab the selected curved arrow and drag around the arc. The model follows orientation calculated from measured joints and body attitude; RBA-1 fingers also follow measured opening. EE focus follows the measured arm-tip position, and missing pose data shows a waiting message. Clicking or grabbing alone does not move the arm. Moving beyond a small screen dead zone jogs at the existing speed while the pointer remains displaced from the grab point. Returning to the grab point, releasing or canceling the touch requests input stop. Drag distance does not specify travel distance or a target pose. The gesture frame freezes at grab time, so live model motion cannot change its direction. Reversing waits for the stop request to be accepted before starting the other direction. An axis that is difficult to resolve in the current view uses upward drag for + and downward drag for −. X/Roll is red, Y/Pitch green and Z/Yaw blue. The curved arrow indicates positive rotation. Dragging a selected EE handle controls the arm; dragging elsewhere in the 3D view rotates the viewpoint. The full-robot PIP only switches the view and never sends movement. In FC frame, movement arrows at the EE show the controller's level foot-center frame, including its foot-twist correction; in EE frame, movement arrows and rotation axes follow the local arm-tip frame. FC display requires updated ManiControl and Network on the robot. If FC telemetry is missing, the drag model shows a waiting message rather than body-aligned axes; existing button controls remain available. Existing axis and −/+ buttons are available in the Button controls tab. Stop EE input stays visible. Changing axes or modes, switching input tabs, a second touch, leaving the screen or hiding the app cancels current input and pending drag starts. Gripper open and close remain separate hold controls. Fix locks the arm position and Control enables manual control again. Reset requests manual entry again and may move to the ready pose. Stow arm and exit stops held input, waits for XYZ, RPY and gripper stop requests to be accepted, then requests Folding/Home and returns home. Header navigation and in-app back navigation use the same folding request. While connected, a failed request keeps the screen open; this does not wait for physical folding completion. On disconnect, pending inputs and queued manual commands are canceled and the screen returns to the hub without automatic folding. A notice states that arm stop and stow are unconfirmed. Reconnection does not resume control: inspect the robot and explicitly start manual control again. This does not guarantee robot-side stopping or completed folding. Closing or reloading the browser tab cannot guarantee this sequence.
For translation, choose FC frame (the existing foot-center movement) or EE frame. In EE frame, +X approaches along the gripper's forward direction and −X retreats. X/Y/Z arrows follow the EE orientation; both axis dragging and the −/+ buttons use that frame. Each new hold captures the measured EE axis, maintains the current orientation reference, and requests translation at the selected fraction of the existing 0.10 m/s jog speed. Changing frames stops the current input and does not start a new movement. Rotation always uses the EE frame. The EE approach stops as a whole at the existing workspace or ground boundary; physical travel still depends on the controller and joint limits. Align the gripper yourself first: object recognition, automatic alignment, contact detection and grasping are not provided. This feature requires the updated ManiControl on the robot as well as the updated GUI; updating the GUI alone does not enable it.
The top action row also contains Robot Control Mode and Chicken Head Mode, beside Reset, Fix, Control and Stow arm and exit. Start manual control before using either mode. Both require arm control to be running and the position lock released, and stop any held EE input before sending their command. Selecting a mode does not confirm that the robot completed the change — check its status and logs.
Manual control uses the existing EE-based pitch tracking without additional automatic base-height adjustment.
In manual control, open EE speed beside Input feedback to set translation and rotation separately from 10% to 100%, or choose Fine, Normal or Fast. The settings apply to axis dragging and −/+ buttons and are saved on this device. Opening settings or changing speed stops the current input; closing the window does not resume movement. These percentages set the requested speed, not a measured speed or travel distance. If speed support is unavailable, update the robot software and app together. The 3D view follows the current FC heading while retaining the measured arm and EE orientation.
Everything about what you see and hear lives on the right.
3D is the source.SLAM — LiDAR robots only. Opens the SLAM / Nav popover and switches the view to the SLAM map.Talk. Sounds picks a sound and plays it on a loop on the robot. Capture asks whether you want a Photo or a Recording; while recording, the button shows the elapsed time in red and tapping it again stops. When the screen is too narrow to fit them beside the center gauges, the two buttons stack above Talk. See Media.Drive, Level and Reset. Pick the view from the drop-down at the top center; progress is shown below it. See the simulator notes in Connection for levels and views.WARNING
With P2G on, the whole video area is a command surface — every tap on the floor moves the robot. Turn it off when you are done.
Driving from the keyboard — on a PC with no gamepad, the keyboard works alongside the virtual joysticks. It works only on this screen: the same keys do nothing on the Hub or in settings.
| Key | What it does |
|---|---|
| W A S D | Left stick — forward/back and sideways |
| Q E | Right stick left / right — turn left / right |
| ↑ ↓ ← → | Right stick |
The robot moves while you hold a key and stops when you let go. The keyboard has no emergency-stop key — E-STOP is the button at the top right of the screen.
While connected, the bottom-left corner shows the current gait name as a small HUD.
On short screens — phones, or a window less than 500 px tall — the layout stays the same and only the pieces get smaller. The joysticks, tool buttons, and walkie-talkie tuck into the corners, and the left card takes only the space above them and scrolls. On narrow screens the cards get narrower too. Nothing is taken away.
On very small screens — less than 720 px wide, such as a small phone in landscape — the screen shows less. The tool buttons at the bottom and the buttons beside Talk show icons only, E-STOP at the top right shows only its emergency icon, and the signal bars in the transfer-rate chip and the gyro at the center shrink or go away. Everything still works the same.
What developer mode adds here is collected on Developer mode.