Pilot City Ideation : Qualcomm Navigation Optimization Project

[SanPranav]

⬆️Click to Return to Table of Contents

Qualcomm Autonomous Vehicle Navigation Enhancement Project

Project Overview

We're building a machine learning application that uses Poway's Open Data Portal transport data to make autonomous vehicles navigate better. Our goal is to cut travel time by 20% for local commuters through traffic prediction, route optimization, and smart speed control.

Scope → Community → Team → Research → Ideate

Who / What / Initial Concept and Idea

• Who: Qualcomm's autonomous vehicle team, Poway city commuters, transportation authorities
• What: ML-powered navigation system using local transport data
• Initial Concept: Create a system that learns from historical and real-time traffic data to predict congestion patterns and optimize routes for autonomous vehicles

Key Questions to Answer:

1. What transport data is available from Poway's Open Data Portal?
2. How can we process this data to create useful navigation insights?
3. What machine learning approaches will work best for traffic prediction?
4. How will users interact with the system?

Establish User Stories / Visuals of UIs

User Stories:

1. As a commuter, I want my autonomous vehicle to automatically take the fastest route to work based on predicted traffic conditions
2. As a city planner, I want to see aggregate traffic flow data to identify bottlenecks
3. As a Qualcomm engineer, I want an interface to monitor system performance and accuracy
4. As a driver, I want to receive alerts about unexpected traffic changes
5. As a fleet manager, I want to optimize routes for multiple vehicles simultaneously

UI Mockups Needed:

• Driver dashboard showing route, ETA, and traffic conditions
• Traffic prediction visualization interface
• System administration panel
• Mobile app interface for remote monitoring

API Endpoints that Correspond to User Stories

Proposed Endpoints:

1. /api/predict-traffic - GET - Returns traffic predictions for specified routes and times
2. /api/optimize-route - POST - Takes start/end points and returns optimized route
3. /api/traffic-data - GET - Returns historical and real-time traffic data
4. /api/system-metrics - GET - Returns system performance metrics
5. /api/vehicle-status - GET/POST - Gets/updates vehicle location and status
6. /api/user-preferences - GET/PUT - Gets/updates user route preferences

Database Model / Draw.io Diagrams to Support APIs

Database Tables:

1. users - Store user profiles and preferences
2. vehicles - Store vehicle information and capabilities
3. traffic_data - Store historical traffic information
4. routes - Store common routes and their properties
5. predictions - Store traffic predictions
6. system_logs - Store system performance metrics

Relationships:

• Users have many vehicles
• Routes have many traffic_data points
• Predictions reference traffic_data
• Draw.io diagram needed to visualize these relationships

Machine Learning or Other Key Technical Features

ML Components:

1. Traffic Flow Prediction Model:

   • Time series forecasting using historical patterns
   • Weather impact correlation analysis
   • Special event traffic pattern recognition

2. Route Optimization Algorithm:

   • Dynamic path finding based on predicted conditions
   • Multi-parameter optimization (time, distance, fuel efficiency)
   • Learning from past route effectiveness

3. Adaptive Speed Control System:

   • Speed recommendation based on traffic density
   • Traffic light timing prediction
   • Safe following distance calculation

Technical Requirements:

• Real-time data processing pipeline
• Edge computing on vehicle hardware
• Cloud-based model training infrastructure
• Data visualization components

Repository Preparations (1 Point)

• Fork from our Smart City Framework repository
• Already includes:
  • User authentication and profile management
  • Blog functionality for project updates
  • Basic data visualization components
  • REST API framework

Additional Components to Add:

• Connect to Qualcomm's vehicle simulation environment
• Set up CI/CD pipeline for regular deployment
• Configure data ingestion from Poway Open Data Portal
• Add Gemini API integration for natural language traffic updates

Titanic to Pilot City Tinker (1 Point)

0.80 Part 1

• Take our existing Titanic survival prediction model framework
• Adapt it to predict traffic congestion instead of survival
• Connect frontend UI where users can input route parameters
• Create "Would your route be congested?" interactive tool
• Use transfer learning to adapt the prediction model

0.90 or Greater, Part 2

• Integrate real Poway traffic data from Pilot City dataset
• Develop complete API structure for traffic prediction
• Connect to autonomous vehicle simulation environment
• Create frontend with meaningful purpose: "Will your commute be faster with our system?"
• Implement A/B testing framework to compare against conventional navigation
• Add visualization showing "before and after" route comparisons

Action Items

1. Set up project repository with initial structure
2. Collect and analyze sample data from Poway Open Data Portal
3. Create initial ML model prototype based on Titanic framework
4. Design database schema and API endpoints
5. Build basic UI mockups for frontend
6. Schedule weekly team sync meetings
7. Establish metrics for measuring the 20% time reduction goal

Team Assignments

• Data Science Lead: Rohan Bojja
• Backend Developer: Aarush Gowda
• Frontend Developer: Vasanth Rajasekaran
• ML Engineer: Nikhil Maturi
• QA/Testing: Nathatn Tejidor
• Project Manager: Pranav Santhosh

⬆️Click to Return to Table of Contents