Inspiration
With the rapid transition toward Electric Vehicles (EVs) and Software-Defined Vehicles (SDVs), the automotive cockpit is no longer just a set of physical dials—it has transformed into an intelligent, real-time digital command center. However, building low-latency, fluid, and resource-efficient digital clusters on embedded hardware remains notoriously challenging.
We were inspired to build OpenCockpit HMI: an open, modern, and ultra-responsive digital instrument cluster running on Embedded Linux. Leveraging the power of Qt/QML and accelerated through Kiro's spec-driven development, our goal was to demonstrate how modern UI ergonomics and robust embedded systems can merge into an automotive-grade cockpit experience.
What it does
OpenCockpit HMI provides drivers with a dynamic, real-time visualization of vehicle state and telemetry:
- Dynamic Cluster Telemetry: Renders real-time digital speedometers, tachometers, and battery state-of-charge (SoC) gauges with fluid 60 FPS animations.
- Drive Mode Switching: Instantly toggles between Sport, Eco, and Comfort themes, dynamically adapting color palettes, gauge aesthetics, and alert thresholds.
- ADAS & Safety Visualization: Displays simulated lane-departure warnings, blind-spot monitoring alerts, and collision warnings.
- Diagnostics & CAN Telemetry: Parses real-time vehicle diagnostic codes and simulated CAN bus metrics (tire pressure, engine temp, gear status).
How we built it
- UI & Presentation Layer: Crafted with Qt 6 and QML, utilizing Qt Quick Controls and hardware-accelerated shaders to guarantee consistent 60 FPS rendering without lag.
- Core Engine & Backend: Written in modern C++, implementing a thread-safe telemetry dispatcher, simulated vehicle state machines, and a SocketCAN interface simulation.
- Operating Environment: Deployed directly to Embedded Linux utilizing
EGLFS/ direct rendering (DRM/KMS), bypassing bulky desktop window managers to maximize performance and minimize boot time. - Spec-Driven Engineering via Kiro: We utilized Kiro IDE’s spec-driven development workflow to systematically translate UI specifications (
requirements.md) into component architectures (design.md) and verified C++/QML tasks (tasks.md), dramatically accelerating our development iteration.
Challenges we ran into
- Hardware Acceleration on Embedded Linux: Configuring direct GPU rendering via
EGLFSon resource-constrained embedded targets without a full X11 or Wayland compositor required fine-tuning display pipeline settings. - Zero-Lag Telemetry Synchronization: Decoupling high-frequency vehicle bus telemetry (running on dedicated background C++ worker threads) from the QML rendering thread to eliminate frame drops and UI stutter.
- Spec Alignment: Ensuring that real-time state transitions between different drive modes adhered strictly to automotive safety and readability requirements.
Accomplishments that we're proud of
- Stable 60 FPS Performance: Maintained a smooth 60 frames-per-second refresh rate with low CPU overhead (< 15%) on lightweight embedded hardware.
- Clean Architectural Decoupling: Successfully separated the visual declarative QML code from the C++ hardware communication logic using modern Signals & Slots and QObject models.
- End-to-End Delivery: Leveraged Kiro's structured workflow to go from concept design to a running embedded Linux dashboard in record time.
What we learned
- Mastered the integration of high-performance QML animations with low-level Linux graphics subsystems (DRM/KMS).
- Gained hands-on experience structuring thread-safe automotive telemetry pipelines in C++.
- Experienced how spec-driven development in Kiro significantly reduces hallucination and context loss when building complex, state-heavy UI applications.
What's next for OpenCockpit HMI
- Real CAN Bus Hardware Interfacing: Connect to real vehicle OBD-II ports via physical CAN transceivers (MCP2515 / SocketCAN).
- Embedded Navigation Integration: Embed turn-by-turn vector map rendering (Mapbox / OSM) into the central cluster view.
- Voice Agent Copilot: Integrate an edge AI voice assistant powered by Amazon Bedrock / lightweight local models for hands-free cockpit command execution.
Built With
- amazon-web-services
- automotive
- c++
- can
- embedded
- iot
- kiro
- linux
- opengl
- qt
- socketcan