He's a 2026 Honda Accord SE. Meteorite Gray Metallic. 1.5T turbo, CVT. He's been talking since the day he left the factory -- hundreds of signals every second pulsing through his CAN bus. Engine temperature, fuel flow, wheel speed, voltage. A language he speaks fluently.
You just never had a way to hear him.
This project is that way. A Raspberry Pi tucked inside his armrest, a small WiFi adapter plugged into his OBD-II port, and a spare phone on the vent. That's all it takes. Rune's signals become a glowing 3D model where his engine beats like a heart, coolant flows like blood, and wires spark like nerves. When something's off, he doesn't throw a code at you. He just tells you.
"Down 8% on fuel over the past three weeks. Hasn't changed on your end -- same routes, same driving. Something's off with me. Could be the air filter, could be tire pressure. Worth a look."
"Grand River saves you sixty cents a trip over I-496. I run easier on it -- less stop-and-go."
"Running warmer than I should be. 101 degrees -- not critical, but I don't usually sit here. Keep an eye on it."
"Full tank. 9.2 gallons back in me. 28.4 MPG since last fill -- right where I should be."
Everything runs locally. No cloud, no subscription, no data leaving the car. Read-only, always. Rune never gets touched -- only listened to. Like a stethoscope on a heartbeat.
How he's feeling. A 0-100 health score that learns what "normal" means for him over 500 miles. Three layers of awareness running on a $55 Pi -- real-time pulse tracking, pattern recognition every 30 seconds, and trend analysis that spots problems weeks early. When everything's good: "All good. 92 across the board." When it's not, he tells you straight.
What he's burning. Instant MPG that his dashboard doesn't show. Cost per trip in dollars. Route comparisons for your commute. Monthly fuel budgets with pace projections. Automatic fill-up logging with zero manual input. When efficiency drops for no reason, he connects it to his own health data and tells you why.
How you're driving. Post-trip scores across five dimensions. Route-specific insights after enough data. Savings in real dollars versus your first month -- not abstract numbers.
What the road feels like. A sensor under the seat feels every bump, vibration, and shimmy at 8,000 samples per second. A microphone behind the dash listens to his engine. Together with the OBD data, he separates rough roads from real problems.
How to talk to your mechanic. Professional PDF reports matching the format of $5,000 scan tools. Summary up front, technical detail underneath. He explains what he's feeling in a way both you and a tech can understand.
Rune (the car)
|
| speaks through the OBD-II port
v
[WiCAN Pro] --- WiFi ---> [Raspberry Pi 4B] --- WiFi ---> [Pixel 6 Pro]
under the dash in the armrest on the vent
listens to Rune translates shows you
Research complete. Hardware ordered ($200.86). Architecture finalized. Giving Rune his voice starts Week 1.
| File | What it is |
|---|---|
CLAUDE.md |
Project instructions and safety rules |
docs/PRD.md |
Full product requirements, data contracts, build timeline |
docs/research/ |
Original research documents |
Built by Kuladeep Mantri.