// PROJECT 005

AI Powered Vehicle Assistant

Edge and cloud AI platform delivering real-time, contextual recommendations to drivers. NVIDIA Jetson at the edge runs local inference for latency-critical decisions while AWS handles fleet analytics and LLM-based personalization — unified over MQTT messaging.

NVIDIA JetsonMQTTAWS IoT Core LLM IntegrationEdge AIPython / C++ TensorFlow LiteOpenCV
// IN-VEHICLE Camera GPS/IMU CAN Bus OBD-II Microphone // JETSON ORIN EDGE INFERENCE ENGINE TFLite / CUDA Inference LOCAL MODELS Driver Monitor Object Detect Route Optimizer Speech NLU MQTT Publisher → AWS // MQTT BROKER AWS IoT Core TLS 1.3 / QoS 1 // AWS CLOUD Lambda (Rules Engine) SageMaker (Retraining) DynamoDB (Trip Data) S3 (Model Registry) // LLM LAYER GPT-4o / Claude API // PERSONALIZATION Driver Profile Store Route History OTA Model Push OTA Updates
Problem
No Contextual Intelligence
Drivers lacked real-time, personalized recommendations based on route, vehicle state, and driver behaviour — all decisions were static or rule-based.
Solution
Edge + Cloud AI Platform
NVIDIA Jetson handles low-latency local inference. AWS cloud manages fleet analytics, LLM personalization, and OTA model updates — all connected via MQTT over TLS.
Result
Real-Time Recommendations
Sub-200ms contextual driver recommendations, personalized route coaching, fatigue detection, and proactive maintenance alerts delivered on-device.
<200msInference Latency
94%Driver Monitor Accuracy
TLS 1.3MQTT Security
OTAModel Updates

System Architecture Overview

The platform operates across two compute tiers. The edge tier (Jetson Orin) runs continuous inference on camera, audio, and CAN telemetry streams — making decisions that require low latency, such as drowsiness detection and emergency braking alerts. The cloud tier (AWS) handles long-horizon analytics: driver profiling, route history, model retraining via SageMaker, and LLM-powered natural language coaching delivered back to the in-vehicle HMI.

All communication between tiers uses MQTT over TLS 1.3 via AWS IoT Core. The Jetson publishes telemetry topics at configurable rates (100 ms–5 s depending on data class) and subscribes to control topics for OTA model payloads and personalization config pushes.

Edge Inference Pipeline (Jetson Orin)

MQTT Topic Architecture

vehicle/{id}/telemetrySpeed, RPM, fuel, GPS — published every 1s, QoS 0
vehicle/{id}/dmsDrowsiness score, gaze vector — published every 200ms, QoS 1
vehicle/{id}/alertsSafety events (fatigue, harsh braking) — retained, QoS 2
cloud/{id}/recommendLLM coaching messages from AWS Lambda → HMI display
ota/{id}/modelModel update payloads from S3 → Jetson OTA agent
cloud/{id}/profilePersonalized driver config pushed from DynamoDB

LLM Integration — Cloud Layer

Vehicle telemetry, trip history, and driver behaviour scores are aggregated in DynamoDB and passed as structured context to a GPT-4o / Claude API call via AWS Lambda. The LLM generates natural-language coaching messages ("You've been driving for 3 hours — rest stop in 12 km") that are routed back to the Jetson HMI display via MQTT.

// Lambda: LLM context builder
const context = {
  driver_fatigue_score: 68,
  trip_duration_hrs: 2.9,
  harsh_events_last_30min: 3,
  fuel_efficiency_trend: "declining",
  next_rest_stop_km: 12,
  driver_profile: "commuter_aggressive"
};

const prompt = `You are an in-vehicle AI assistant.
Driver context: ${JSON.stringify(context)}
Generate a concise, actionable recommendation (max 20 words).`;

const response = await openai.chat.completions.create({
  model: "gpt-4o",
  messages: [{ role: "user", content: prompt }],
  max_tokens: 40
});
// → "You seem fatigued. A rest stop is 12 km ahead. Consider taking a break."

OTA Model Update Pipeline

Technology Stack

Edge HardwareNVIDIA Jetson Orin NX (16 GB, 100 TOPS)
Edge RuntimeTensorRT, TFLite, OpenCV 4.x, Python 3.10 + C++ bridge
MessagingMQTT 5.0, AWS IoT Core, Paho MQTT client, TLS 1.3
CloudAWS Lambda, SageMaker, DynamoDB, S3, IoT Rules Engine
LLMOpenAI GPT-4o API / Anthropic Claude API (via Lambda)
Vehicle InterfaceCAN 2.0B @ 500 kbps, OBD-II PID reader, GPIO HMI

Security Considerations

← All Projects