RIG-Hover

RIG-Hover, also known as Wenwen, is an open-source desktop robotics hardware platform from the RIG family. It combines an ultra-narrow two-wheel self-balancing chassis, an independently actuated head, a round expressive display, voice AI, camera input, and the RIG-Omni ESP32-S3 firmware stack.

RIG-Hover is designed for makers, robotics developers, educators, and open hardware communities who want more than a static desktop gadget. It is printable, buildable, tunable, hackable, and expressive.

Demo video:

Bilibili BV1qP756mEUY

What Is RIG-Hover?

RIG-Hover is an open-source self-balancing desktop robot kit built around ESP32-S3 and the RIG-Omni firmware architecture. It is a compact hardware development platform for learning and experimenting with embedded robotics, balance control, voice AI, expressive animation, camera tools, and maker-friendly mechanical design.

The project is intentionally shaped like a global open-source DIY kit: printable mechanical parts, documented hardware modules, reproducible firmware, structured tuning guidance, and clear extension points for developers.

Key Features

Ultra-narrow self-balancing chassis: two feedback FOC hub motors create a compact balancing platform suitable for desktop motion.

Low center of gravity: the battery and electronics are stacked close to the wheel axis to improve dynamic stability.

Head-body decoupling: the base can move or rotate while the head servo looks independently left or right.

Expressive round display: a 240 x 240 GC9A01 round LCD plays EAF/LVGL facial animations.

Voice and AI interaction: microphone, speaker, Wi-Fi, wake word, cloud ASR/TTS, and RIG-Omni application services.

Camera-ready platform: GC0308 camera integration for snapshots and future vision experiments.

Developer-friendly firmware: ESP-IDF, FreeRTOS tasks, MCP tools, Web tuning API, OTA-ready architecture.

Hardware Specifications

Item	Specification
MCU	ESP32-S3-WROOM-1, dual-core 240 MHz
Motion system	2 FOC hub motors with encoder feedback
Head actuator	1 EM3 serial bus servo
Degrees of freedom	3: differential wheels + horizontal head rotation
Display	1.09 inch 240 x 240 round TFT, GC9A01
Camera	GC0308, DVP interface
Audio	MEMS microphone + 2 W speaker, I2S
IMU	QMI8658C
Network	BluFi provisioning, 2.4 GHz Wi-Fi
Battery	2S 7.4 V 500 mAh lithium battery
Size	About 61 mm x 61 mm x 139 mm
Weight	About 375 g
Max glide speed	About 10 cm/s
Runtime	About 1.5 h balancing standby, about 1 h head-body interaction

Documentation

Document	Purpose
docs/02-Documentation-Map.md	Reading paths and document index
docs/03-Quick-Start.md	Power-on, Wi-Fi setup, and basic interaction
docs/04-Bill-of-Materials.md	Electronics, printed parts, fasteners, and sourcing rules
docs/05-Replication-Guide.md	End-to-end reproduction workflow
docs/06-Assembly-Guide.md	Mechanical assembly and wiring checks
docs/07-Firmware-Guide.md	RIG-Omni build, flash, board selection, and source map
docs/08-Tuning-Guide.md	Balance tuning, Web debug API, and safety boundaries
docs/09-Developer-Guide.md	Architecture, tasks, MCP, and extension patterns
docs/10-FAQ.md	Troubleshooting
11-Contributing.md	Contribution guide

Firmware Source

RIG-Hover firmware is part of LuwuDynamics/rig_omni, an open-source ESP32-S3 multi-robot firmware framework.

Important Hover files:

main/boards/hover/hover_board.cc: board startup, LCD, buttons, camera, UART, IMU, MCP tools.

main/boards/hover/xgo.cc: servo protocol, wheel feedback, LQR balance control, BLE/XGO command parsing.

main/boards/hover/hover_debug_server.cc: LAN Web tuning page and /api/data, /api/setendpoints.

main/boards/hover/board_config.h: display, camera, IMU, touch, and UART pin definitions.

Build From Source

git clone https://github.com/LuwuDynamics/rig_omni.git

cd rig_omni

source ~/esp/esp-idf/export.sh

idf.py set-target esp32s3

idf.py menuconfig

Select:

RIG-Omni -> Board Type -> RIG-Hover

Then build, flash, and monitor:

idf.py build

idf.py -p /dev/ttyUSB0 flash monitor

Windows port example:

idf.py -p COM5 flash monitor

Safety

RIG-Hover is a self-balancing robot. Motor direction, encoder direction, IMU orientation, zero angle, battery voltage, loose mechanics, or incorrect tuning can cause falling or violent oscillation.

Hold the robot during first power-on and first tuning.

Test suspended wheels first, then hand-held balance, then free-standing balance.

Keep fingers, cables, and loose objects away from the wheels.

Do not expose the Web tuning service to the public internet.

Use the specified motor screw length; screws that are too long may damage the motors.

01-Overview