Vehicle Velocity Prediction Using Artificial Neural Networks and - PowerPoint PPT Presentation

21-23 20 Vehicle Velocity Prediction Using Artificial Neural Networks and Effect of Real-World Signals on Prediction Window Tushar D. Gaikwad , Zachary D. Asher, Aaron Rabinowitz, Farhang Motallebiaraghi, Thomas Bradley, Alvis Fong, Rick Meyer Western Michigan University, Colorado State University

Special Thanks Dr. Thomas H. Bradley 2020-01-0729 2

Special Thanks This material is based upon work supported by the U.S. Department of Energy’s Office of Energy Efficiency and Renewable Energy (EERE). The specific organization overseeing this report is the Vehicle Technologies Office under award number DE-EE0008468. 2020-01-0729 3

Introduction Methods Results Conclusions Agenda: ฀ Introduction/Background ฀ Drive Cycle Development ฀ LSTM ฀ Assessment methods ฀ Effect of different signals ฀ Summary ฀ Results ฀ Future Goal 2020-01-0729 4

Introduction Methods Results Conclusions Intelligent Transportation Systems(ITS) • The shift that we are witnessing toward Intelligent Transportation Systems(ITS), will be the most disruptive since the Initial days of automobiles. • It has potential to completely transform the movement of people and goods, enabling safer and smarter transportation. • ITS consists of several technologies such as Advanced Driver Assistance System(ADAS), Automated Driving Functions(ADF), Vehicle to Vehicle(V2V) and Vehicle to Infrastructure(V2I) communication. 2020-01-0729 5

Introduction Methods Results Conclusions Use of Autonomous Vehicles • Autonomous vehicle technology is the key to improve driver and passenger safety. • Control strategies and AI software that powers it are some of the most critical components. • Increase in computational capabilities, enable to train complex and deep neural networks 2020-01-0729 6

Introduction Methods Results Conclusions Motivation for velocity prediction Fuel economy strategy: Optimal Energy Management Systems Ego Vehicle velocity Predictions determines the constraints of the energy optimization problem to increase fuel economy [Gaikwad, Tushar D., Zachary D. Asher, Kuan Liu, Mike Huang, and Ilya Kolmanovsky. Vehicle Velocity Prediction and Energy Management Strategy Part 2: Integration of Machine Learning Vehicle Velocity Prediction with Optimal Energy Management to Improve Fuel Economy. No. 2019-01-1212. SAE Technical Paper, 2019.] Safety improvement strategy: Collision Risk Estimator Ego Vehicle velocity Predictions can be integrated into collision risk estimators thereby helping drivers avoid accidents. [Phillips, Derek J., Real-time Prediction of Automotive Collision Risk from Monocular Video (2019)] 2020-01-0729 7

Introduction Methods Results Conclusions Novel Contribution • Use of different real-world signals in groups to estimate effect on prediction. • Understanding effects of different inputs in LSTM for varied prediction window. • Use of two assessment methods to understand results. 2020-01-0729 8

Introduction Methods Results Conclusions Vehicle Velocity Prediction Strategy • Drive Input • Deep Learning and Machine Learning Models • Prediction Window • Assessment Model LSTM 2020-01-0729 9

Introduction Methods Results Conclusions Drive Cycle Development and Signal Recording • Collected in August 2019 at Fort Collins, Colorado. • Dataset from repeated drives collected along a fixed route by the same driver. • Route Details 1. Parking Lot 2. West on Mulberry until Shields 3. South on Shields until Prospect 4. East on Prospect until College 5. North on College until Mulberry 6. West on Mulberry until Parking Lot 7. Parking Lot 2020-01-0729 10

Introduction Methods Results Conclusions Data Collection Sr. No. Parameters 1 Time • Autonomous driver assistance system 2 Latitude (ADAS) data for the vehicle forward cone 3 Longitude from smart radar. 4 Distance • Vehicle to infrastructure (V2I) data in the form 5 Brake Pedal Position of traffic signal information and segment 6 Transmission Gear travel times. 7 Engine Speed • EGO Vehicle Parameters using Freematics 8 Max Torque logger. 9 Min Torque 10 Engine Torque 11 Turn Signal 12 Vehicle Speed 13 Acceleration Longitude 14 Acceleration Latitude 15 Yaw Rate 16 Altitude 17 SPaT 18 Segment Speed 19 ADAS 2020-01-0729 11

Introduction Methods Results Conclusions Recorded Signals Radar Object Detections immediately Velocity on the Longitude vs Latitude in front. vs time 2020-01-0729 12

Introduction Methods Results Conclusions Recorded Signals Velocity vs Time for one Velocity on the map Terrain drive instance. 2020-01-0729 13

Introduction Methods Results Conclusions LSTM(Long Short-term memory) • Special type of Recurrent Neural Network (RNN) • A typical LSTM unit is composed of a cell, an input gate, an output gate, and a forget gate • The cell remembers values and the three gates regulate the flow of information into and out of the cell. • Forget Gate: Decides what information to discard from the cell. • Input Gate: Decides which values from the input to update the memory state. • Output Gate: Decides what to output based on input and the memory of the cell 2020-01-0729 14

Introduction Methods Results Conclusions Assessment methods 2019-01-1212 15

Introduction Methods Results Conclusions Assessment methods Increase in • MAE = 1.6763 m/s • MAE = 1.6763 m/s MAE and Time Increase • Time Shift = 10 Seconds shift • Time Shift = 0 Seconds in MAE 2019-01-1212 16

Introduction Methods Results Conclusions Assessment methods • MAE = 4.3901 m/s • Time Shift = 10 Seconds 2019-01-1212 17

Introduction Methods Results Conclusions Effect of Different Signals on Prediction Current Velocity Current Velocity Current Velocity Data Group A GPS GPS GPS Previous 5 Seconds Group B Previous 5 Seconds Data Group D Group E EGO and Engine Data Group F EGO and Engine Current Velocity Parameters Group A Parameters GPS SPat SPat Data Group B Previous 5 Seconds Segment Speed EGO and Engine Radar Parameters Current Velocity Current Velocity GPS Current Velocity GPS Previous 5 Seconds Group D GPS Group B Data Group E EGO and Engine Previous 5 Seconds Group B Data Group G Previous 5 Seconds Parameters Data Group C EGO and Engine SPat EGO and Engine Parameters Parameters Segment Speed Segment Speed Radar Data 2019-01-1212 18

Introduction Methods Results Conclusions Effect of Different Signals on Prediction 2019-01-1212 19

Introduction Methods Results Conclusions Effect of Different Signals on Prediction 2019-01-1212 20

Introduction Methods Results Conclusions Effect of Different Signals on Prediction Prediction with Group E Prediction with Group D 2019-01-1212 21

Introduction Methods Results Conclusions Effect of Different Signals on Prediction Window 2019-01-1212 22

Introduction Methods Results Conclusions Forward Prediction Window Prediction for every second shown for Every 10 second interval 2019-01-1212 23

Introduction Methods Results Conclusions Forward Prediction Window Prediction for every second shown for Every 10 second interval 2019-01-1212 24

Introduction Methods Results Conclusions Summary • Autonomous vehicles is the key to move towards ITS • It uses AI to power different strategies to enable safer and smarter transportation. • Velocity prediction is very important to develop those strategies. Group D LSTM 10 second MAE and Time Shift prediction Model MAE = 1.78 m/s Time Shift =2.35 LSTM 2019-01-1212 25

Introduction Methods Results Conclusions Conclusion • For Evaluating Results its better to use MAE along with time shift. • Dataset with SPaT data performs better in terms of getting accurate results • LSTM performs the best in terms of MAE, which is 1.78m/s, at prediction window of 10 seconds. • Increase in Prediction Window Increases MAE and Time shift in case of ANN. Future Work • Study with More drive cycle on different types of roads • Inclusion of V2V data and Camera data • More V2I data. • Implementation of trained model on NVIDIA drive PX2 to test prediction accuracy on the road. 2019-01-1212 26

Thank you Name Tushar D. Gaikwad , Zachary D. Asher, Aaron Rabinowitz, Farhang Motallebiaraghi, Thomas Bradley, Alvis Fong, Rick Meyer University/Work Western Michigan University, Colorado State University Address 4601 Campus Dr, Kalamazoo, MI 49008 Email tushar.d.gaikwad@wmich.edu / zach.asher@wmich.edu 2019-01-1212 27