180 likes | 318 Views
High Resolution In Vehicle Sensing. Nagui M. Rouphail Director, ITRE Professor of Civil Engineering NC State University. DriveSense14 October 30-31, 2014. The challenge. 1 billion highway vehicles SAFETY 1.2 million traffic fatalities per year ENERGY 30% of world Energy EMISSIONS
E N D
High Resolution In Vehicle Sensing Nagui M. Rouphail Director, ITRE Professor of Civil Engineering NC State University DriveSense14 October 30-31, 2014
The challenge • 1 billion highway vehicles • SAFETY • 1.2 million traffic fatalities per year • ENERGY • 30% of world Energy • EMISSIONS • 25% of world CO2 Emissions • TRAFFIC • 1.5 hours per day on a vehicle
Outline Description of in-vehicle sensor Data description and demonstration Research questions and hypotheses Planned capabilities (VIV)
In-Vehicle Sensor: Background Partnership with TUL(Technical University of Lisbon) and ITds (software company in Lisbon) Funded collaboration through NSF international supplement for a just concluded NSF award Sensor developed in Portugal by ITds and TUL through an Innovation co-fund award Initial prototype was to provide feedback to driver on fuel use and emissions via a secure website Ongoing prototype testing through funding from the University of Maryland National UTC
The In-Vehicle Sensor i2D INTELLIGENCE TO DRIVE
How it Works GPRS/GSM
All available PIDs from OBD as: speed (odometer), rpm, engine temperature, accelerator position, error codes, VIN… (most of them on a 1 Hz basis) • From additional sensors: location (GPS), 3 axis accelerometer (up to 50 Hz local), altitude (barometer) … • fuel consumption (i2D algorithms), CO2 and other pollutant emissions, engine cold temperature points, slope, distances, driving periods, driving events (Stops, predefined alerts over speed, rpm, accelerations…), average speed, energy efficiency for each trip… • trip mapping and reconstruction, benchmarking, driving indicators, driving learning support, Driving Profiling, why and where are you spending fuel, … The Data Levels (min 1Hz Resolution) 1st Level Raw Data 2nd Level Processed Data
Database at NC State • ~2 million records of data seconds are collected each month, each having 40 data fields from about 10-15 vehicles driven by student/ staff volunteers • About ~3,500 miles of travel (low use) • Consumes ~200MB of memory (xlsx format) • Available to NC State in a SINGLE table format • Hard to perform queries, changes, etc. • NCSU broke down the table into several tables connected to each other in a SINGLE database • More efficient query and search • Is needed to perform faster visualization
Other Databases • Individual users and fleet managers access website https://app.i2d.co/ • Basic configuration of unit, vehicle, password • User friendlyreports, visualizations, etc • Research Website http://research.i2d.co • Simple web access for researchers to download raw data • i2D public website https://www.i2d.co/i2dpubportal/login.xvw • Shows the overall performance of drivers and vehicles anonymously • NCSU Website under construction http://www.redconverge.com/i2d • Based on SQL database • Performs faster search and visualization
Private Driver Website View (1) Events…
Private Driver Website View (2) Trip Summaries, benchmarking and fuel waste reports
1stLevel communication Channel 2ndLevel communication Channel • SAFETY Applications • Queue Warning • “Event / Exception Trigger Based” alert; each generating “n” customized messages that are automatically delivered by the system to identified vehicles • Preventing accidents and traffic jams PLANNED CAPABILITY –December 2014 VIV – Vehicle to Infrastructure to Vehicle • Data justfor VIV purposes: • SAFETY Applications • TRAFFIC Applications • Datamay be associated with • Fleet Mangmt. • Individual usage • UBI (insurance) M2M DATA M2M DATA i2D Dual Communication System (M2M) for VIV M2M communication establishes an IP connection 2 independent, parallel, communication channels are created a Random ID is generated for each trip –> No Privacy issues Priority Real Time
Research Questions / Hypotheses Generating driving profiles from Hi Res data Developing micro-scale vehicle interaction models based on driver profiles (car-following, lane changing, gap acceptance) Testing hypotheses of micro-scale driver behavior vs. long term safety record Distinguishing contributing factors to crashes (driver behavior, road/ traffic control effects) Testing impact of feedback on eco-driving perform. Long term driving trends vs. economic factors
Research Questions / Hypotheses Real time data quality checks and imputations Testing regional and national travel demand model route choice assumptions (UE vs. SE vs. SO) Testing assumptions about traffic signal timing Testing the value of and compliance with travel information to calibrate/ validate ATIS models Feasibility of PHYD or PAYD tolling schemes Privacy issues…
Questions Thank you !