Recent Projects
Heavy Truck Active Safety Systems
Integrated Vehicle-Based Safety Systems (IVBSS) Field Operational Test for Eaton Corporation (2009)Developed tools and algorithms to analyze the large-volume IVBSS test database. The goal of the research was to identify salient data trends, patterns and anomalies and to create/validate data models and warning algorithms.
Performed data mining, data analysis, and data modeling to identify causes of high incidence of crash warning alerts detected during the IVBSS Field Operational Test. Identified sources and causes of crash warning alerts in terms of: equipment design/performance failures, environmental conditions and driver behavior. Recommended corrective actions, and monitored vehicle performance after adjustment to determine whether faults had been corrected.
Created an SQL database tool for the registration of all trips of vehicles passing through designamted map coordinates. This mechanism provided a framework for developing performance profiles of a variety of vehicles operating in similar road geometries.
Transportation Modeling
Model Autonomous Driving Functionality in the U.S. Department of Transportation's TRANSIMS (Transportation Analysis and Simulation) System (2008 - 2009)Vehicular route guidance and autonomous driving are distinct but related transportation technologies that are likely to be deployed jointly in the future in order to provide navigation capabilities to driverless vehicles. The purpose of this research project is to examine how these two technologies work together and to assess the potential benefits and challenges involved.
While fully autonomous driving systems are still in their infancy, many limited autonomous driving mechanisms, such as lane-keeping, cruise control and crash avoidance systems have been widely and successfully deployed. These systems provide partial control of the vehicle for limited periods, with a human driver supervising the activation, deactivation and operation of the autonomous system. This project addresses a limited autonomous driving function, lane-keeping. The research is thus aimed at providing answers to the following questions:
* Can limited autonomous driving and route guidance technologies function together to produce beneficial results?
* What are the required elements of the guidance and autonomous driving components that are required for a successful marriage of the two technologies?
* Are there any features of the two components that that may be able to operate independently but not together, and if so, what accommodations can be devised? For example, is there any aspect of lane keeping that would prevent the successful execution of a turn or lane change, while the lane keeping function is engaged?
* What kind of system controls and interface between the two component technologies would be most useful and easy to use by human operators?
* What new insights into major problems associated with the two component technologies can be provided by this research?
* How can the lessons to be learned in applying route guidance to limited autonomous driving help in the application of route guidance to fully automated driving?
For the purposes of this research, the TRANSIMS system provides route guidance functionality with modifications implemented by Cognometrics to model lane-keeping functionality in selected TRANSIMS vehicles. The performance of a TRANSIMS system that has been modified in this way is compared to the performance of an unmodified baseline system for a variety of lane-keeping parameters.
Intelligent Vehicle Systems
Perform independent research for Cognometrics and specific studies for the General Motors Corporation to investigate and implement smart car technologies and systems (2006 - 2008)
1. Parking Lot Algorithm Developed an intelligent agent that determines whether a vehicle is traveling in a parking lot. Machine learning is employed to train the system to recognize driving patterns and behavior that typically occur in parking lots.
2. Driver Eye Gaze Classifier Developed a Driver Eye Gaze Classifier that determines, with a high degree of accuracy, whether a driver's eyes are on or off the road ahead based on driver input (e.g. steering, braking, throttle, etc.). Machine learning is used to train the on-board system to recognize driving patterns associated with eye gaze positions of typical drivers. Accurate assessment of driver gaze is an important ITS component and is an indicator of driver distraction that can be used to trigger a driver alert warning. Knowledge of driver gaze patterns can be used to improve the design of vehicle displays and controls.
3. Driver Braking Pattern Analysis Performed a study and analysis of driver braking patterns to determine the possible need for system modifications to improve performance and reliability in the most commonly occuring braking situations.
4. Curved Driving Analysis Performed a study of curved driving to assess the impact on vehicle speed and acceleration of driving on road segments of varying radii of curvature.
5. Road Width Inference Researched the feasibility of using GPS vehicle position data together with map databases to infer the width of road segments by computing the distances of moving vehicles from the center of the road segment as identified on the map.
Business Intelligence
Provide Business Intelligence and Customer Relationship Management Services to Amway Corporation (2005)1. Lifetime Customer Value (LTV) Analysis Developed a probabilistic model of selected company market segments using Markov Chain Models. The model provided a capability to estimate the Customer Lifetime Value (LTV) by market segment.
2. Marketing Campaign Analysis Analyzed the effectiveness of promotion and marketing campaigns using propensity models such as:
* Target vs. Control group comparisons
* Loyalty Value
* Pre-Post program comparisons
3. Customer Analytics Developed and implemented product performance analytics and measures to assess the success of the company's various products in the marketplace. Using pattern recognition techniques, evaluated performance in such areas as:
* Sales by major product line and Store Keeping Unit (SKU)
* Repeat purchase patterns over time, including:
Average order amount
Average time between purchases
* Brand loyalty and the tendency of purchasers to defect
(switch subsequent purchases to competitive brands)
4. Data Mining and Database Balancing Performed database balancing to ensure consistency between the database of one of Amway's Latin American affiliates and the company's forecasting and history databases, its business intelligence reporting system (Microstrategy) and its global data warehouse.
5. IT System Implementation Developed a standard Customer Relationship Management reporting template to measure customer sales and retention activities for one of the companies major Asian markets.
Automotive Crash avoidance
Assist General Motors in developing, testing and implementing an automotive crash avoidance system (2004 - 2005)1. Data Mining and Analysis Performed data analysis and data mining on a database of test data accumulated during field testing of an Automotive Forward Crash Warning System installed in a fleet of test vehicles. Data collected from on-board radar and camera devices, GPS (Global Positioning Satellite) and path-prediction map databases was evaluated and analyzed through a variety of tools such as SQL queries, MATLAB Statistical Analysis Toolbox functions and m-file programs, and in-house developed tools to display sensor data in real-time.
2. Driver Distraction Analysis Performed a study to determine the state of the vehicle operator (the driver) from vehicle data such as steering, braking and yaw rate. Used pattern recognition and SQL/Matlab time-sequence analysis to cluster vehicle input data that correlated to specific patterns of distracted driver behavior and the occurrence of warnings.
3. Crash warning classification and cluster analysis Performed crash warning clustering studies to understand the causes and circumstances surrounding false crash warning alerts. For example, identified the circumstances in which certain kinds of driving behavior (e.g. curved driving) were most likely to create false crash warning alerts. Also identified physical locations in which certain kinds of crash warning alerts were most likely to occur.
4. Tailgating Study Conducted an analysis of tailgating data to provide a framework for developing alert warning thresholds. Defined a tailgating metric and compiled tailgating data over driving time intervals during which drivers operated within various tailgating thresholds.
5. Database Mapping and Documentation Developed a data dictionary of crash warning data that maps data collected from on-board sensors to data elements in an SQL database.
IT Training
Present an on-going series of inhouse developed IT IT training courses and seminars focussing on data mining, information retrieval, data modeling and business intelligence.
| Powered by webRider Technologies |
| Copyright © 2003 Cognometrics, Inc., all rights reserved |