Hardware Deployment Guide

NVIDIA Orin Nano · Raspberry Pi · RISC-V · Docker

Deploy NanoSolana on real hardware using the build-tag HAL. One binary runs identically on a $50 Pi or a $500 Orin Nano — or silently stubs on macOS for development.

HAL Architecture · Supported Boards · Orin Nano · Raspberry Pi · Wiring · Docker

🔩 Hardware Abstraction Layer (HAL)

NanoSolana uses a build-tag HAL so the same binary runs on real hardware or in simulation. Build tags are automatic — go build on Linux links hal_linux.go, everywhere else gets hal_stub.go.

HAL Stack 
┌────────────────────────────────────────────────┐
│         MawdBot Agent (OODA Loop)               │
├────────────────────────────────────────────────┤
│          DeviceManager (pkg/hardware)           │
│  Modulino drivers · Event routing · Mapping    │
├────────────────────────────────────────────────┤
│             HAL Interface (internal/hal)        │
│   Bus · Device · GPIO · PWM · Platform         │
├──────────────────┬─────────────────────────────┤
│  hal_linux.go    │     hal_stub.go             │
│  (periph.io)     │     (synthetic data)        │
│  Real I2C/GPIO   │     Logs + mock sensors     │
│  Orin/RPi/RISCV  │     macOS/Windows/CI        │
└──────────────────┴─────────────────────────────┘
🐧

hal_linux.go

  • periph.io I2C driver
  • Real GPIO + PWM
  • Orin / RPi / RISC-V
  • /dev/i2c-1 direct access
🧪

hal_stub.go

  • Synthetic sensor data
  • Log-only GPIO
  • macOS / Windows / CI
  • No hardware needed
⚙️

Build Commands

  • make build — macOS stub
  • make orin — NVIDIA Orin
  • make rpi — Raspberry Pi
  • make riscv — RISC-V

📋 Supported Boards

Board Arch Build Target I2C GPIO Notes
NVIDIA Orin Nano arm64 make orin ✓ ✓ Primary target, /dev/i2c-1
NVIDIA Orin NX arm64 make orin ✓ ✓ Same as Nano
Raspberry Pi 5 arm64 make rpi ✓ ✓ /dev/i2c-1
Raspberry Pi 4B arm64 make rpi ✓ ✓ /dev/i2c-1
Raspberry Pi Zero 2W arm64 make rpi ✓ ✓ Low RAM, works fine
Raspberry Pi Zero W armv6 make rpi-armv6 ✓ ✓ Limited CPU
StarFive VisionFive 2 riscv64 make riscv ✓ ✓ Experimental
BeagleBone AI-64 arm64 make orin ✓ ✓ TI TDA4VM
x86 Server amd64 make linux-amd64 ✗ ✗ Software-only
Mac (dev) arm64/amd64 make build ✗ ✗ Stub mode

🟢 NVIDIA Orin Nano Setup

1

Flash & Boot

# Flash JetPack 6.x via SDK Manager or SD card image
# Boot into Ubuntu desktop, connect via SSH
2

Enable I2C

# Check I2C buses
ls /dev/i2c-*

# If not present, enable in device tree
sudo /opt/nvidia/jetson-io/jetson-io.py
# Select "Configure Jetson 40pin Header" → Enable I2C

# Install tools
sudo apt install -y i2c-tools

# Test — you should see your Modulino addresses
sudo i2cdetect -y 1
3

Deploy NanoSolana

# On your dev machine:
make orin
scp build/mawdbot-orin build/mawdbot-tui-orin user@orin-nano:~/

# On the Orin Nano:
chmod +x ~/mawdbot-orin ~/mawdbot-tui-orin
sudo mv mawdbot-orin /usr/local/bin/mawdbot
sudo mv mawdbot-tui-orin /usr/local/bin/mawdbot-tui

# Add user to i2c group (logout/login after)
sudo usermod -aG i2c $USER

# Initialize, configure, scan
mawdbot onboard
cp ~/.mawdbot/.env.example ~/.mawdbot/.env
nano ~/.mawdbot/.env

mawdbot hardware scan
mawdbot hardware test
4

Systemd Service

sudo tee /etc/systemd/system/mawdbot.service << 'EOF'
[Unit]
Description=MawdBot Solana Trading Agent
After=network-online.target
Wants=network-online.target

[Service]
Type=simple
User=mawdbot
Group=i2c
EnvironmentFile=/home/mawdbot/.mawdbot/.env
ExecStart=/usr/local/bin/mawdbot ooda --interval 60
Restart=always
RestartSec=10

# Security
NoNewPrivileges=true
ProtectSystem=strict
ReadWritePaths=/home/mawdbot/.mawdbot
PrivateTmp=true

# Device access
SupplementaryGroups=i2c
DeviceAllow=/dev/i2c-1 rw

[Install]
WantedBy=multi-user.target
EOF

sudo systemctl daemon-reload
sudo systemctl enable mawdbot
sudo systemctl start mawdbot
sudo journalctl -u mawdbot -f

🍓 Raspberry Pi Setup

1

Enable I2C

sudo raspi-config
# Interface Options → I2C → Enable

sudo apt install -y i2c-tools
sudo i2cdetect -y 1
2

Deploy

# On your dev machine:
make rpi
scp build/mawdbot-rpi pi@raspberrypi:~/mawdbot
3

Wiring (Qwiic)

Connect Modulino nodes to the RPi's I2C pins:

SDAGPIO 2 — Pin 3
SCLGPIO 3 — Pin 5
3.3VPin 1
GNDPin 6

Or use a Qwiic HAT / breakout for plug-and-play daisy-chaining.

🔌 Arduino Modulino® Wiring

All Modulino nodes connect via Qwiic (I2C daisy-chain). No soldering required.

Qwiic daisy-chain 
[Orin Nano / RPi]
     │
     │ Qwiic Cable
     ▼
┌──────────┐   Qwiic   ┌──────────┐   Qwiic   ┌──────────┐
│ Distance │ ─────────► │ Movement │ ─────────► │  Thermo  │
│  0x29    │            │  0x6A    │            │  0x44    │
└──────────┘            └──────────┘            └──────────┘
                                                      │
     ┌────────────────────────────────────────────────┘
     │ Qwiic
     ▼
┌──────────┐   Qwiic   ┌──────────┐   Qwiic   ┌──────────┐
│  Pixels  │ ─────────► │  Buzzer  │ ─────────► │ Buttons  │
│  0x6C    │            │  0x3C    │            │  0x7C    │
└──────────┘            └──────────┘            └──────────┘
     │ Qwiic
     ▼
┌──────────┐
│   Knob   │
│  0x76    │
└──────────┘

Hardware ↔ Agent Mapping

SensorAgent Integration
Distance 0x29 Proximity alert — pause trades if someone is near the hardware
Movement 0x6A Vibration/tilt detection — emergency stop on physical disturbance
Thermo 0x44 Thermal throttle — reduce activity if board overheating
Knob 0x76 Live strategy tuning — maps to position_size, stop_loss, or RSI threshold
Buttons 0x7C A = force cycle · B = toggle simulate · C = checkpoint vault
Pixels 0x6C OODA phase display (teal→purple→amber→green) + signal strength bar
Buzzer 0x3C Trade alerts (rising tone) · error alerts (low tone)

🐳 Docker with Hardware Access

1

Build Image

docker build -t mawdbot .
2

Run with I2C

docker run --rm \
  --device /dev/i2c-1 \
  --env-file .env \
  mawdbot ooda --interval 60
3

Full Hardware Access

docker run --rm \
  --privileged \
  --env-file .env \
  mawdbot ooda --interval 60

📊 Monitoring

monitoring commands 
# Live sensor stream
$ mawdbot hardware monitor

# System status
$ mawdbot status

# Logs (systemd)
$ journalctl -u mawdbot -f --no-pager

# Resource usage (should be <10MB RSS)
$ ps aux | grep mawdbot
💾

Resource Budget

  • Binary: ~9.6 MB
  • RAM: <10 MB RSS
  • Boot: <1 second
  • CPU: idle ~0%
🔒

systemd Security

  • NoNewPrivileges
  • ProtectSystem=strict
  • PrivateTmp
  • DeviceAllow /dev/i2c-1
🌡️

Sensor Thresholds

  • Proximity <5cm — wake cycle
  • Tilt detected — pause trades
  • Temp overheating — throttle
  • Knob turn — adjust RSI

Ready to deploy on hardware?

View on GitHub Full Guide