Hardware — M5Stack StackChan¶

TL;DR¶

The robot body is the M5Stack StackChan kit: an M5Stack CoreS3 (ESP32-S3) head on a 2-servo chassis.
The CoreS3 supplies the SoC, display, camera, mic array, speaker, IMU, proximity, microSD, NFC, IR — all integrated.
The StackChan kit adds the head-yaw servo, head-pitch servo, 12 RGB LEDs, 3-zone touch panel, 700 mAh supplementary battery, USB-C, and the 3D-printed body.
Firmware on the device is built from m5stack/StackChan — an Arduino C++ codebase that bundles the XiaoZhi AI agent client. It is not the same codebase as meganetaaan/stack-chan (the original Moddable/JS project) or 78/xiaozhi-esp32 (generic voice-assistant firmware).
The device advertises itself over the Xiaozhi WebSocket protocol and exposes on-device tools via MCP (see protocols.md).

The SoC and board: M5Stack CoreS3¶

All values from docs.m5stack.com/en/core/CoreS3 (see references.md).

Component	Spec
SoC	ESP32-S3, dual-core Xtensa LX7 @ 240 MHz
Flash	16 MB
PSRAM	8 MB Quad
Display	2.0″ IPS, 320×240, ILI9342C, capacitive touch
Camera	GC0308, 0.3 MP (built-in)
Proximity / ambient-light	LTR-553ALS-WA
IMU	BMI270 (6-axis accel + gyro)
Magnetometer	BMM150 (3-axis) — gives 9-axis combined with BMI270
Mic codec	ES7210, dual-mic input
Speaker amp	AW88298, 16-bit I2S, 1 W
PMU	AXP2101
Battery (internal)	500 mAh Li-ion
RTC	BM8563
microSD	Supported, up to 16 GB
Wi-Fi	2.4 GHz only
BLE	Yes
USB	USB-C (device + power)
Dimensions	54.0 × 54.0 × 15.5 mm (unit only)
Weight	72.7 g (unit only)

What the StackChan kit adds on top¶

Values from m5stack/StackChan README (see references.md):

Component	Spec
Head-yaw servo	Feedback servo, 360° horizontal rotation
Head-pitch servo	Feedback servo, 90° vertical movement
Front-panel LEDs	12 × RGB, arranged in two rows
Touch panel	3-zone (beyond the display's own touch)
NFC	Yes, reader + writer
IR	Transmitter + receiver
Supplementary battery	700 mAh
Chassis	3D-printed body, base, feet (STL published)

Note on battery size. The CoreS3's internal battery is 500 mAh; the StackChan kit documents 700 mAh. The StackChan kit appears to bundle an external cell that supersedes or supplements the CoreS3 internal one. Check the physical kit to confirm before quoting battery life numbers.

Firmware lineage¶

Three related codebases — do not confuse them:

Repo	Language	Purpose	Runs on StackChan?
`meganetaaan/stack-chan`	TypeScript / JavaScript on Moddable SDK	Original open-source Stack-chan (Shinya Ishikawa)	Yes (but not what we run)
`m5stack/StackChan`	Arduino C++	M5Stack's official firmware — bundles XiaoZhi AI agent, targets CoreS3	Yes — this is what we flash
`78/xiaozhi-esp32`	Arduino C++	Generic multi-board voice assistant firmware (70+ target boards)	Runs on the same ESP32-S3 but is a different application — you pick one, not both

Our pipeline uses m5stack/StackChan because it comes with the robot-body integration (servos, avatar rendering, LED patterns, MCP tools mapped to peripherals) already done. 78/xiaozhi-esp32 is the upstream protocol reference — the voice channel speaks the same WebSocket protocol regardless.

On-device MCP tools¶

The device acts as an MCP server — after the WS hello handshake, it advertises its tools to xiaozhi-server via tools/list (JSON-RPC 2.0 inside type: mcp messages). See protocols.md for the exact wire format.

Tool names follow the dotted-namespace convention from the 78/xiaozhi-esp32 MCP protocol doc (e.g. self.audio_speaker.set_volume, self.get_device_status). The registration sites in the firmware use McpServer::AddTool for public tools and McpServer::AddUserOnlyTool for privileged/hidden ones.

Per internal deployment observation, the live firmware advertises 11 tools. The mapping below is from that observation plus the m5stack/StackChan README's feature list — verify against the handshake logs (docker logs xiaozhi-esp32-server | grep tools/list) before relying on exact tool names:

#	Tool (functional)	Hardware touched
1	Head yaw	Yaw feedback servo
2	Head pitch	Pitch feedback servo
3	LED color	12× RGB LEDs
4	Camera — `take_photo`	GC0308 camera
5	Reminders / timer	RTC (BM8563) + software
6	Volume	AW88298 amp
7	Display brightness	ILI9342C backlight
8	Screen theme	Avatar renderer
9	Face expression	Avatar renderer (see protocols.md)
10	Get device status	All (battery, RSSI, uptime)
11	Reboot	MCU

Action item to make this table canonical: capture a real tools/list response and commit the tool-name column verbatim. Tracked in latent-capabilities.md as an observability gap.

Peripherals the firmware could expose but doesn't (per current observation)¶

These are real hardware features with no documented MCP tool in the default firmware today. See latent-capabilities.md for prioritization.

Peripheral	Capability	Why it'd matter
BMI270 + BMM150 (9-axis IMU)	Shake / gesture / orientation detection	Tap-to-activate, shake-to-reset; orientation-aware responses
LTR-553 proximity sensor	Hand-approach detection, ambient light	Wake-on-approach; auto-dim at night
NFC module	Tag read/write	Tap an NFC card/toy to trigger a scripted interaction
IR tx/rx	Learn + replay IR codes	Universal-remote mode for legacy appliances
microSD slot	Offline asset storage	Pre-bundled sound packs, offline fallback voices
3-zone touch panel	Multi-zone tap/swipe	Gesture controls without using the display's touch
Camera (beyond `take_photo`)	Video streaming / on-device vision preprocessing	Privacy-preserving local vision before sending to a VLLM

Safety-relevant hardware facts¶

Mic is I2S via ES7210. Hot whenever the firmware chooses — there is no hardware mic-mute. The privacy-indicator LED item in ROADMAP.md exists because of this.
Servos can move fast. Feedback servos in a kids' environment can startle. The StackChan kit uses the M5Stack Avatar library's ease functions; the velocity cap is a firmware-side choice, not a hardware limit. See the "Servo speed caps" item in ROADMAP.md.
Camera has no shutter. Software-only enable. The take_photo MCP tool should always co-activate a distinct LED state (see child-safety task).