This is the webpage for NSF Forgetful Database project.
Background
This project seeks implementation-proven answers to the question: when do data expire? The proposed work is motivated by a cyber-physical transportation application – fleets of connected and autonomous vehicles (CAVs) driving on potentially icy roads, where safety-critical road friction information is shared via a wireless data link to a central database that mediates data averaging. The insight of our approach is to treat the database as a sensor to the vehicle, and the aggregation of data within the database as a sensor-averaging process. This abstraction reveals that the database functionality can be formally modeled to quantify the impacts of timeliness as well as uncertainty of information, enabling analysis of the stability of the vehicle’s behavior and the stability of the vehicle/database interaction as a function of the averaging process
Objectives
In V2x systems, databases operate in feedback loops and can enter “locked-up” conditions :
- We want databases to store and share data, but need them to forget “old” dat
- We want vehicles to measure data, but each vehicle needs data recorded by others.
- We want the vehicles to maneuver enough to create accurate data, but need vehicles to be stable.
Main Goals
The major goal of this project is to design data collection and storage systems that are intentionally "forgetful" by design, such that data is retrieved (remembered) and shared in safety-critical situations only within the time and location regions for which they apply. Recognizing that data from the database may be used for decision-making by the very agents contributing data to the database, we are developing tools that study this interactive behavior. We specifically are designing algorithms to discover trends that determine how long the data from sensors is relevant, both for individual sensors and when information is obtained by groups of sensors. We demonstrate the foundational knowledge through a database system interacting with connected vehicles to estimate road conditions and use this information in vehicle safety systems to prevent vehicles from skidding on wet or icy roads.
The technical goals of the project are: to create a methodology for gathering, storing, and processing large quantities of data created by multiple similar dynamic physical systems by considering data quality; to use contextual information about the gathered database information to enhance the quality of the output data and maximize database performance; to generate a scalable database querying methodology that predicts measurements for feedforward control of agents based on dynamically updating estimates of data quality; to create an analysis approach for determining the stability of feedback control when some or all of the feedback signal is obtained from the database; to analyze the stability and performance of the database itself, including the ability of the database to request queries from the agents populated data within the database.
Team Members
Project PIs
Sean Brennan
Professor
Mechanical Engineering
The Pennsylvania State University
Craig Beal
Associate Professor
Mechanical Engineering
Bucknell University
Cindy Chen
Associate Professor
Computer Science
UMass Lowell
Kshitij Jerath
Assistant Professor
Mechanical Engineering
UMass Lowell
Students
Liming Gao
Mechanical Engineering
The Pennsylvania State University
Hossein Haeri
Mechanical Engineering
UMass Lowell
Niket R Kathiriya
Computer Science
UMass Lowell
Satya Prasad Maddipatla
Mechanical Engineering
The Pennsylvania State University
Juliette Faith Mitrovich
Mechanical Engineering
The Pennsylvania State University
Rinith Pakala
Computer Science
UMass Lowell
Alumni
Wushuang Bai
Mechanical Engineering
The Pennsylvania State University
Eric Fan
Computer Science
UMass Lowell
Usha Sravani Ganta
Computer Science
UMass Lowell
Lorina Sinanaj
Computer Science
UMass Lowell
Achievements
Detection of surface friction conditions from a fleet of vehicles and the use of the aggregated data for safe operation of these vehicles
For more details
How it works
The solution includes:
- Use Allan Variance (AVAR) to dynamically identify meaning of “old” data and optimal averaging windows.
- Develop idea of context to organize and discover properties of data that further refine variance.
- Organize database systems for optimal Allan Variance calculations.
- Test results via synchronization of regional-level traffic simulators with chassis simulators, and with a steer-by-wire instrumented vehicle.
- Fast variance codebase with 10^4 speed improvement.
- Proved variance method gives optimal averaging windows.
- Developed analytical solution for minimum preview necessary for chassis control with changing friction.
- Designed and tested database organization structures to support AVAR.
- Developed tools to encompass regions of influence (ROI) for perturbation analysis of vehicle impacts.
Code repositories
- Public Repo for Forgetful DataBases work
- VehicleSimulations_ProjectExamples_NSFForgetfulDatabases
- FDB_AVAR_Based_Algorithms
- forgetfulDBs
Vehicle information
Publication list
Published Peer-reviewed Papers- Sinanaj, Lorina; Haeri, Hossein; Maddipatla, Satya Prasad; Gao, Liming; Pakala, Rinith; Kathiriya, Niket; Beal, Craig; Brennan, Sean; Chen, Cindy; Jerath, Kshitij; “Granulation of Large Temporal Databases: An Allan Variance Approach,” SN Computer Science, VOL 4, 2022, doi: 10.1007/s42979-022-01397-2.pdf
- Gao, Liming; Beal, Craig; Mitrovich, Juliette; Brennan, Sean; “Vehicle Model Predictive Trajectory Tracking Control with Curvature and Friction Preview,” in 10th Annual IFAC Advances in Automotive Control Symposium, 28-31 August, 2022, Columbus, Ohio, USA, https://doi.org/10.1016/j.ifacol.2022.10.288.pdf
- Bai, Wushuang; Maddipatla, Satya Prasad; Pelletier, Evan; Gao, Liming; Brennan, Sean; “Determining Region of Influence of Ego-Vehicle on Roadways for Vehicle Decision Making,” in 2022 Road Safety and Simulation International Conference (RSS), 08-10 June, 2022, Athens, Greece. pdf (No DOI)
- Gao, Liming; Beal, Craig; Bai, Wushuang; Maddipatla, Satya Prasad; Chen, Cindy; Jerath, Kshitij; Haeri, Hossein; Sinanaj, Lorina; Brennan, Sean; “Boxes-based Representation and Data Sharing of Road Surface Friction for CAVs,” in 2022 Road Safety and Simulation International Conference (RSS), 08-10 June, 2022, Athens, Greece. pdf (No DOI)
- Sinanaj, Lorina; Haeri, Hossein; Gao, Liming; Maddipatla, Satya Prasad; Chen, Cindy; Jerath, Kshitij; Beal, Craig; Brennan, Sean; “Allan Variance-based Granulation Technique for Large Temporal Databases,” in 13th International Conference on Knowledge Management and Information Systems (KMIS), October 25-27, 2021.pdf(No DOI)
- Maddipatla, Satya Prasad; Haeri, Hossein; Jerath, Kshitij; Brennan, Sean; “Fast Allan Variance (FAVAR) and Dynamic Fast Allan Variance (D-FAVAR) Algorithms for both Regularly and Irregularly Sampled Data,” in 2021 Modeling, Estimation and Control Conference (MECC), Austin, TX, October 24-27, 2021, https://doi.org/10.1016/j.ifacol.2021.11.148.pdf
- Gao, Liming; Bai, Wushuang; Leary, Robert; Varadarajan, Krishna; Brennan, Sean; “ROS Integration of External Vehicle Motion Simulations with an AIMSUN Traffic Simulator as a Tool to Assess CAV Impacts on Traffic,” in 2021 Modeling, Estimation and Control Conference (MECC), Austin, TX, October 24-27, 2021, https://doi.org/10.1016/j.ifacol.2021.11.281.pdf (It is concurrent with the project but not supported by NSF.) .
- Gao, Liming; Beal, Craig; Fescenmyer,Daniel; Brennan, Sean; “Analytical Longitudinal Speed Planning for CAVs with Previewed Road Geometry and Friction Constraints,” in 2021 IEEE International Intelligent Transportation Systems Conference (ITSC), September 19-22, 2021,Indianapolis, IN, United States, 2021, https://doi.org/10.1109/ITSC48978.2021.9564602.pdf
- Gao, Liming; Maddipatla, Satya Prasad; Beal, Craig; Jerath, Kshitij; Chen, Cindy; Sinanaj, Lorina; Haeri, Hossein; Brennan, Sean; “A Micro-simulation Framework for Studying CAVs Behavior and Control Utilizing a Traffic Simulator, Chassis Simulation, and a Shared Roadway Friction Database,” 2021 American Control Conference (ACC), 2021, pp. 1650-1655, doi: 10.23919/ACC50511.2021.9483221. pdf
- Haeri, Hossein; Beal, Craig; Jerath, Kshitij; "Near-Optimal Moving Average Estimation at Characteristic Timescales: An Allan Variance Approach," in IEEE Control Systems Letters, vol. 5, no. 5, pp. 1531-1536, Nov. 2021, doi: 10.1109/LCSYS.2020.3040111. pdf
- Maddipatla, Satya Prasad; Brennan, Sean; "Synchronization and Feedback Loop Integration of a Non-real Time Microscopic Traffic Simulation with a Real-time Driving Simulator using Model-Based Prediction," 2021 American Control Conference (ACC), 2021, pp. 4376-4382, doi: 10.23919/ACC50511.2021.9482970.pdf
- Beal, Craig; Brennan, Sean; “Friction detection from stationary steering manoeuvres,” Veh. Syst. Dyn., vol. 58, no. 11, pp. 1736–1765, 2020,doi: 10.1080/00423114.2019.1645862. pdf
- Beal, Craig; Brennan, Sean; “Modeling and friction estimation for automotive steering torque at very low speeds,” Veh. Syst. Dyn., vol. 3114, 2020, doi: 10.1080/00423114.2019.1708416. pdf
- Bai, Wushuang; Borek, John; Gao, Liming; Vermillion, Chris; Brennan, Sean; "Determining the Region of Influence of a Signalized Traffic Intersection by Analysis of Heavy-duty Diesel Vehicle Fuel Consumption," 2021 American Control Conference (ACC), 2021, pp. 1867-1874, doi: 10.23919/ACC50511.2021.9482959.pdf
-
Gao, Liming; Mitrovich, Juliette; Beal, Craig; Bai, Wushuang; Maddipatla, Satya Prasad; Chen, Cindy; Jerath, Kshitij; Haeri, Hossein; Sinanaj, Lorina; Brennan, Sean; “Boxes-based Representation and Data Sharing of Road Surface Friction for CAVs,” Data Sci. Transp. pdf
- Pakala, Rinith; "Distributed Edge Computing System Setup for Vehicle Communication" in 2023 DATA Conference
- Mitrovich, Juliette; Maddipatla, Satya Prasad; Gao, Liming; Guler, Ilgin; Beal, Craig; Brennan, Sean; "Analysis of Friction Utilization Within a Roadway Network Using Simulated Vehicle Trajectories" in 2023 IEEE Conference on Control Technology and Applications (CCTA), IEEE, Aug. 2023
- Mitrovich, Juliette; Maddipatla, Satya Prasad; Gao, Liming; Beal, Craig; Brennan, Sean; "Synthesizing Feasible Vehicle Trajectories from Microscopic Traffic Simulations" in 2023 IEEE Conference on Control Technology and Applications (CCTA), IEEE, Aug. 2023
- Haeri, Hossein; "Stream Learning Across Multiple Scales: Assessing Model Stability in Presence of Continual Concept Drift," in 2023 IEEE Transactions on Knowledge and Data Engineering
- Haeri, Hossein; Kathiriya, Niket "Adaptive Granulation: Data Aggregation at the Database Level," in 2023 International Conference on Database Systems for Advanced Applications (DASFAA)
- Pakala, Rinith; “DISTRIBUTED EDGE COMPUTING SYSTEM FOR VEHICLE COMMUNICATION,” M.S. Thesis, Department of Computer Science, University of Massachusetts Lowell, April 2023.
- Mitrovich, Juliette; “DATABASE-MEDIATED NETWORK FRICTION UTILIZATION ANALYSIS TO IMPROVE ROAD SAFETY,” M.S Thesis, Department of Mechanical Engineering, The Pennsylvania State University, March 2023.
- Gao, Liming; “DATABASE-MEDIATED PREVIEW OF ROADWAY FRICTION AND MODEL PREDICTIVE PATH TRACKING CONTROL FOR CONNECTED VEHICLES,” Ph.D. Thesis, Department of Mechanical Engineering, The Pennsylvania State University, October 2022. pdf
- Sinanaj, Lorina; “ALLAN VARIANCE-BASED GRANULATION TECHNIQUE FOR LARGE TEMPORAL DATABASES,” M.S. Thesis, Department of Computer Science, University of Massachusetts Lowell, May 2021. pdf
Images and Media
-
NSF 2021 CPS_LightningTalk
-
MECC 2021: Fast AVAR Algorithms
-
MECC 2021: ROS Integration with Aimsun
-
ITSC 2021: Analytical Longitudinal Speed Planning for CAVs with Previewed Road Geometry and Friction Constraints
-
ACC 2021: A Micro-simulation Framework for Studying CAVs Behavior and Control
-
ACC 2021: Near Optimal Moving Average Estimation at Characteristic Timescales
-
Allan Variance as a Statistic for Change Detection.
-
Vehicle Model Predictive Path Tracking Control with Geometry and Friction Preview - Update.
-
Vehicle Longitudinal Speed Planning with Previewed Road Geometry and Friction Constraints.
-
Manual Implementation of R* Tree.
-
Traffic in State College and the colors represent the friction coefficient.
-
Traffic on I99 near State College.
-
Traffic in downtown, State College.
-
Traffic in State College.
Simulated datasets
- DATASET 1: Friction changes only with time. It varies from 0.9 to 0.2 and back to 0.9 after every 350s as shown below.
- DATASET 2: The simulated road of length 2000m has bridges at stations (200, 205), (455, 465), (765, 780), (980, 990), (1240, 1245), (1495, 1505), (1705, 1710), and (1910, 1920).
As shown below, friction on bridges varies between 0.9, 0.4, and 0.2 every 500s.
Friction on the remaining road parts varies between 0.9, 0.7, and 0.4 every 500s, as shown below.
Broader Impacts:
The broader impacts of the proposed work are:
- This work can save lives and suffering by improving safety-critical decision-making used by CAVs driving under dangerous operational situations.
- The methods and algorithms developed in this work can guide the deployment of all database systems wherein trust in data quality versus timeliness of information are competing factors: healthcare, transportation, news systems and social communication, food quality and inspection systems, etc.
- It broadens participation in our research community via a workshop for practicing system engineers on Allan variance and on database stability/performance as it relates to feedback control; participation of college-level students via a distributed undergrad/grad curriculum for senior and early graduate-level systems engineers that connects database systems to mechatronics and automation; and outreach to incoming students, namely the recruitment of new engineers through summer camp activities. quality and inspection systems, etc.
Funding
This project is supported by the National Science Foundation under grant numbers CNS-1932509, CNS-1931927, CNS-1932138 “CPS: Medium: Collaborative Research: Automated Discovery of Data Validity for Safety-Critical Feedback Control in a Population of Connected Vehicles”.