A lean Rust microservice for the Seeed MR60BHA2 60 GHz mmWave radar sensor on Linux.

---

Note

This project talks directly to the bare MR60BHA2 radar module over UART — either via the Raspberry Pi's GPIO serial port or through a USB-to-TTL adapter on any Linux PC. No XIAO ESP32C6 or ESPHome stack required. (The Seeed kit ships with a XIAO running ESPHome as a Wi-Fi bridge; this daemon replaces that entire stack with a single native binary.)

The daemon reads the sensor over UART and streams aggregated sensor data as JSON-Lines over a Unix domain socket. The protocol library is no_std-compatible.

Architecture

flowchart TD
    SENSOR(Seeed MR60BHA2 sensor)
    SENSOR -->|UART 115200 baud| MR60BHA2D(mr60bha2d daemon)
    MR60BHA2D --> JOURNALD("journald (tracing)")
    MR60BHA2D -->|Unix Socket: /run/mr60bha2/radar.sock| JSON("JSON-Lines stream (~8 Hz) to any connected client")

Loading

The MR60BHA2 emits several different frame types at ~8 Hz (targets, vitals, presence). The daemon aggregates them into a unified snapshot emitted on a configurable timer.

Crates

Crate	Description
mr60bha2-proto	Protocol library: zero-alloc frame parser (state machine). no_std-compatible.
mr60bha2d	Streaming daemon: reads UART, aggregates frames, broadcasts JSON over Unix socket.

Sensor Data Format

The daemon emits one JSON line per snapshot (~8 Hz). Uses SI units throughout.

{
  "ts": 1744489123.456,
  "targets": [
    {
      "x": 0.42,
      "y": 1.71,
      "speed": -0.17,
      "dist": 1.76,
      "angle": 13.8
    }
  ],
  "vitals": {
    "heart_rate": 72.0,
    "breath_rate": 16.0,
    "heart_phase": 0.31,
    "breath_phase": 1.02
  },
  "presence": true
}

Field	Unit	Description
ts	seconds (Unix)	Timestamp
x	metres	Horizontal position (+ right, − left of sensor)
y	metres	Distance in front of sensor (always positive)
speed	m/s	Radial speed (+ approaching, − receding)
dist	metres	Euclidean distance from sensor
angle	degrees	Angle from boresight
vitals.heart_rate	bpm	Heart rate
vitals.breath_rate	bpm	Breathing rate
vitals.heart_phase	rad	Heart waveform phase
vitals.breath_phase	rad	Breathing waveform phase
presence	bool	Human presence detected

vitals is omitted until the sensor emits at least one vitals frame. Up to 3 targets can be tracked simultaneously.

Vitals vs. targets — important distinction

The sensor has two independent operating modes that run in parallel:

Capability	Range	What you get
Target tracking	up to 6 m	Position (x/y), speed, and presence for up to 3 people
Vital signs (heart rate, breathing)	≤ 1.5 m	One global measurement — not per-target

Vitals are designed for a single stationary person (the manufacturer recommends sleep scenarios only). The sensor does not attribute vitals to individual targets — it simply reports one heart rate and one breath rate for whoever is closest. When vitals mode is active (person within 1.5 m), presence-detection sensitivity may decrease.

Quick Start

Requirements

• Rust 1.74+ (or install via rustup)
• A serial port connected to a Seeed MR60BHA2 sensor (direct UART — no XIAO needed)
• Linux (any architecture; tested on x86_64 and aarch64)
• Sensor firmware v1.6.12 or later recommended (fixes target tracking within 1.5 m). See the Seeed wiki for upgrade instructions.

Build

cargo build --release

Configuration

The daemon works out of the box with sensible defaults. A config file is optional — if none is found, the following defaults are used:

Setting	Default
device	/dev/ttyAMA0
baud_rate	115200
socket_path	/run/mr60bha2/radar.sock
window_ms	125
log_level	info

To override any setting, create a TOML config file:

device = "/dev/ttyUSB0"
baud_rate = 115200
socket_path = "/run/mr60bha2/radar.sock"
window_ms = 125
log_level = "info"

Common serial device paths:

• /dev/ttyAMA0 — Raspberry Pi GPIO UART
• /dev/ttyUSB0 — USB-to-serial adapters
• /dev/ttyACM0 — CDC ACM devices (e.g. XIAO ESP32C6 bridge)

Run

# Run with defaults (no config file needed)
mr60bha2d

# Run with explicit config
mr60bha2d /path/to/config.toml

# Override log level via environment
RUST_LOG=debug mr60bha2d

Connect a client to read the stream:

socat - UNIX-CONNECT:/run/mr60bha2/radar.sock

Deployment

Cross-Compilation

# Install cross (Docker-based cross-compiler)
cargo install cross

# Build for aarch64
cross build --target aarch64-unknown-linux-gnu --release

Or use the provided justfile:

just build                                    # default: aarch64-unknown-linux-gnu
TARGET=x86_64-unknown-linux-gnu just build    # override target

Pre-built Binaries

Pre-built binaries for x86_64 and aarch64 Linux are available on the Releases page.

systemd

Copy the provided unit files to the target system:

cp deploy/mr60bha2d.service /etc/systemd/system/
cp deploy/mr60bha2d.tmpfiles /etc/tmpfiles.d/mr60bha2d.conf

# Create runtime directory
systemd-tmpfiles --create

# Edit the service to match your serial device if needed
# Default: /dev/ttyAMA0
systemctl edit mr60bha2d

# Enable and start
systemctl daemon-reload
systemctl enable --now mr60bha2d

The service unit includes security hardening (ProtectSystem, NoNewPrivileges, PrivateTmp, etc.).

Automated Deployment

# Deploy to a remote host (requires just + cross)
DEPLOY_HOST=user@hostname just deploy

Consuming the Stream

Any process can connect to the Unix socket and receive the JSON-Lines stream:

# Live output
socat - UNIX-CONNECT:/run/mr60bha2/radar.sock

# Pretty-print with jq
socat - UNIX-CONNECT:/run/mr60bha2/radar.sock | jq .

# Python example
import socket, json

sock = socket.socket(socket.AF_UNIX, socket.SOCK_STREAM)
sock.connect("/run/mr60bha2/radar.sock")
for line in sock.makefile():
    snapshot = json.loads(line)
    print(snapshot["targets"], snapshot.get("vitals"), snapshot["presence"])

The JSON format is compatible with radar-dash — a sensor-agnostic web frontend that works with any daemon emitting this schema.

Related Projects

Project	Description
radar-dash	Sensor-agnostic HTML5 radar visualisation dashboard. Works directly with mr60bha2d over WebSocket.
ld2450-rs	Equivalent daemon for the HLK-LD2450 24 GHz sensor — same JSON schema, position + Doppler without vital signs.

Wiring

The MR60BHA2 uses a 5-pin connector. UART pins operate at 3.3V TTL level.

Sensor Pin	Connection
5V	5V power supply
GND	Ground
TX	UART RX on your host
RX	UART TX on your host
IO	(optional, interrupt output)

Raspberry Pi GPIO example (3.3V UART — direct connection, no level shifter needed):

Sensor Pin	Pi Pin
TX	Pin 10 (GPIO15 / UART RX)
RX	Pin 8 (GPIO14 / UART TX)

If using a Raspberry Pi, disable Bluetooth to free the primary UART:

# /boot/firmware/config.txt (or /boot/config.txt on older systems)
enable_uart=1
dtoverlay=disable-bt

Sensor Specs

Parameter	Value
Frequency	60 GHz ISM band
Max detection range	3 m
Detection angle	±60° azimuth
Max targets	3 simultaneous
Data rate	~8 Hz
Interface	UART (3.3V TTL), 115200 baud 8N1
Supply voltage	5V DC
Extra features	Heart rate, breathing rate, presence detection

License

Licensed under either of

• Apache License, Version 2.0
• MIT License

at your option.

Contribution

Unless you explicitly state otherwise, any contribution intentionally submitted for inclusion in the work by you, as defined in the Apache-2.0 license, shall be dual licensed as above, without any additional terms or conditions.