跳到主要內容

臺灣博碩士論文加值系統

(44.200.140.218) 您好!臺灣時間:2024/07/19 01:40
字體大小: 字級放大   字級縮小   預設字形  
回查詢結果 :::

詳目顯示

我願授權國圖
: 
twitterline
研究生:黃柏瑋
研究生(外文):Po-wei Huang
論文名稱:利用遞歸神經網路和注意力機制來預測不同場景下車輛的速度
論文名稱(外文):Vehicle speed prediction with RNN and attention model under different semantics
指導教授:施吉昇
指導教授(外文):Chi-Sheng Shih
口試委員:叢培貴林風
口試委員(外文):Pei-Kuei TsungPhone Lin
口試日期:2018-07-27
學位類別:碩士
校院名稱:國立臺灣大學
系所名稱:資訊網路與多媒體研究所
學門:電算機學門
學類:網路學類
論文種類:學術論文
論文出版年:2018
畢業學年度:106
語文別:英文
論文頁數:39
中文關鍵詞:車輛軌跡速度場景遞歸神經網路注意力機制
相關次數:
  • 被引用被引用:0
  • 點閱點閱:165
  • 評分評分:
  • 下載下載:0
  • 收藏至我的研究室書目清單書目收藏:0
本論文旨在利用預測周遭車輛的速度來達成防衛性駕駛。要達成防衛性駕駛,首先要知道周遭車的意圖和預測其速度和軌跡。但是,既有的預測機制有幾個問題。第一,使用了很多資訊,像是前前車的速度,來預測前車的速度。這樣的資訊,實際上是沒有的。第二,既有的模型無法從資料中抓出子序列的規律。第三,既有的模型通常只鎖定單一模型,而沒有分析在不同路段或駕駛類型的影響。面對這些問題,本文提出了這些改進。第一、使用單台車的資料,來做預測,並探索其適用範圍。第二、利用注意力機制來抓出子序列的規律來提升預測準度。第三、分出不同的場景,如轉彎、直行、換車道,和道路的型態,並找尋模型最適合的範圍。以結果來說,相比於既有卡爾曼濾波的作法,利用遞歸神經網路和注意力機制,本文達到了11\%的提升。此外,透過和多台車模型的比較,驗證單台車模型能適用於直行的預測。此外,在接近自由駕駛時,預測會更佳。
In this work, we focus on vehicle velocity prediction for defensive driving. In order to achieve defensive driving, we need to understand and predict the motion of surrounding cars. However, the existing velocity prediction mechanism has three drawbacks. Firstly, they need too much information for prediction and some information are not accessible actually. Secondly, they failed to extract sub-patterns from the training data and thus may cause inaccuracy. Thirdly, only very few scenarios are considered. To deal with these problems, we start with single car model for prediction and explore its limitation by comparing it with multi-car model. Then, we use attention model to increase the accuracy. Finally, we examine the prediction under different road types and maneuver to broaden our scenario. Overall, with attention mechanism, we achieve a 11\% improvement over existing solution. Moreover, we found that single car model could be used in go straight behavior, and it has best result in free driving periods.
Acknowledgments ii
摘要 iii
Abstract iv
1.Introduction 1
1.1 Motivation . . . . . . . . . . . . . . 1
1.1.1 Motion prediction for defensive driving . 1
1.1.2 Problems of existing prediction mechanism for defensive driving .............................. 2
1.2 Our solution and result . . . . . . . . . . 3
1.3 Contribution . . . . . . . . . . . . . . . . 3
1.4 Thesis Organization . . . . . . . . . . . . 4
2 Background and Related Work. 5
2.1 Background . . . . . . . . . . . . . . . . .. . 5
2.1.1 Data preprocessing . . . . . . . . . . . .. . 5
2.1.2 Prediction model . . . . . . . . . . . . . . .5
2.2 Related Work . . . . . . . . . . . . . . . . . .7
2.2.1 Previous study on motion prediction . . . . . 7
2.2.2 Previous study on time series prediction . . .7
2.2.3 Previous study using road semantics 7
3.System architecture and problem definition 8
3.1 Target problem . . . . . . . . . . . . . . . . .8
3.1.1 Challenge I : Some model used many information that’s actually not available . . . . . . . . . . . . . . .8
3.1.2 Challenge II : Fail to find patterns within subsequence . . . . . .. . . . . .. . . . . .. . . .9
3.1.3 Challenge III : Models not trained and tested on various scenario .. . . . . .. . . . . .. . . . . . 9
3.2.1 Our approach for these problems . . . . . . . 10
3.2.2 Our approach for challenge I and II . . . . . 10
3.2.2 Our approach for challenge III . . . . . . . .11
3.3 Problem Definition . . . . . . . . . . . . . . .11
3.3.1 Motion prediction. . . . . . . . . . . . . . .11
3.3.2 Requirement of motion prediction . . . . . . .12
3.3.3 Explore the limitation of single car model . .12
3.3.4 Effectiveness of single car model under different scenario . . . . 12
3.4 System architecture . . . . . . . . . . 13
4. Design and implementation 15
4.1 Implement existing system architecture for on-board sensing data . . . . . 15
4.2 Dataset . . . . . . . . . . . . . . . . . . . . 16
4.3 From raw data to sliding window of velocity . . 16
4.3.1 Group the dataset by bus id . . . . . . . . . 16
4.3.2 Group the dataset into trajectories by time disjointness . . . . . . 17
4.3.3 Sensor fusion for tangential velocity . . . . 17
4.3.4 Stop-and-go filtering . . . . . . . . . . . . 17
4.3.5 Sliding window retrieval . . . . . . . . . . .18
4.4 Seq2seq models . . . . . . . . . . . . . . . . .18
4.4.1 Baseline : constant velocity . . . . . . . . .18
4.4.2 Baseline : kalman filter . . . . . . . . . . .18
4.4.3 Baseline: Encoder-decoder . . . . . . . . .. .19
4.5 Design of data classifier . . . . . . . . . . 23
4.5.1 Classifier for speed limit . . . . . .. . . . 23
4.5.2 Classifier for maneuver . . . . . . . . . . . 24
4.5.3 Classifier for average speed . . . . . . . . 24
4.5.4 Explore the prediction under different scenarios . . . . . . . . . .. . .. . .. . .. . .. . .. . .. . .. . 25
4.6 Comparison of single car model and multi-car model . . . . . . . . . . . . .. . .. . .. . .. . .. . .. . ..25
4.6.1 NGSIM US-101 . . . . . . . . . . . . . . . . .26
4.6.2 The multi-car model . . . . . . . . . . . . . 26
5 Performance Evaluation . . . . . . . . . . . . . 28
5.1 Experiment Environment . . . . . . . . . . . . .28
5.2 Performance metric . . . . . . . . . . . . . . .28
5.3 Experiment on new model . . . . . . . . . . . . 29
5.3.1 Experiment on private dataset . . . . . . . . 29
5.3.2 Experiment on NGSIM US-101 . . . . . . . . . .30
5.4 Prediction under different scenario . . . . . . 32
5.4.1 Experiment for maneuver class . . . . . . . . 32
5.4.2 Experiment of maneuver for comparison between single car model and multi-car model. . . . . . . . . . . .33
5.4.3 Experiment for speed limit and average speed .35
6.Conclusion 37
Bibliography 38
[1] Florent Altché and Arnaud de La Fortelle. An LSTM network for highway trajectory prediction. CoRR, abs/1801.07962, 2018.
[2] Samer Ammoun and Fawzi Nashashibi. Real time trajectory prediction for collision risk estimation between vehicles. In Intelligent Computer Communication and Processing, 2009. ICCP 2009. IEEE 5th International Conference on, pages 417–422.
IEEE, 2009.
[3] ARTC. Automatic research and testing. https://www.artc.org.tw/
upfiles/ADUpload/knowledge/tw_knowledge_499017376.pdf, 2017.
[4] Dzmitry Bahdanau, Kyunghyun Cho, and Yoshua Bengio. Neural machine translation by jointly learning to align and translate. CoRR, abs/1409.0473, 2014.
[5] Dongyao Chen, Kyong-Tak Cho, Sihui Han, Zhizhuo Jin, and Kang G. Shin. Invisible sensing of vehicle steering with smartphones. In Proceedings of the 13th Annual
International Conference on Mobile Systems, Applications, and Services, MobiSys’15, pages 1–13, New York, NY, USA, 2015. ACM.
[6] Kyunghyun Cho, Bart van Merrienboer, Çaglar Gülçehre, Fethi Bougares, Holger Schwenk, and Yoshua Bengio. Learning phrase representations using RNN encoderdecoder for statistical machine translation. CoRR, abs/1406.1078, 2014.
[7] Mordechai (Muki) Haklay and Patrick Weber. Openstreetmap: User-generated street maps. IEEE Pervasive Computing, 7(4):12–18, October 2008.
[8] Rubén Izquierdo, Ignacio Parra, Jesús Muñoz-Bulnes, D Fernández-Llorca, and MA Sotelo. Vehicle trajectory and lane change prediction using ann and svm classifiers. In Intelligent Transportation Systems (ITSC), 2017 IEEE 20th International
Conference on, pages 1–6. IEEE, 2017.
[9] Xiao-Yun Lu and Alexander Skabardonis. Freeway traffic shockwave analysis: Exploring the ngsim trajectory data. 07 2018.
[10] Minh-Thang Luong, Hieu Pham, and Christopher D. Manning. Effective approaches to attention-based neural machine translation. CoRR, abs/1508.04025, 2015.
[11] J. Nilsson, J. Silvlin, M. Brannstrom, E. Coelingh, and J. Fredriksson. If, when, and how to perform lane change maneuvers on highways. IEEE Intelligent Transportation Systems Magazine, 8(4):68–78, winter 2016.
[12] SeongHyeon Park, Byeongdo Kim, Chang Mook Kang, Chung Choo Chung, and Jun Won Choi. Sequence-to-sequence prediction of vehicle trajectory via LSTM encoder-decoder architecture. CoRR, abs/1802.06338, 2018.
[13] Yao Qin, Dongjin Song, Haifeng Cheng, Wei Cheng, Guofei Jiang, and Garrison W. Cottrell. A dual-stage attention-based recurrent neural network for time series prediction. CoRR, abs/1704.02971, 2017.
[14] Fan Wu, Kun Fu, Yang Wang, Zhibin Xiao, and Xingyu Fu. A spatial-temporalsemantic neural network algorithm for location prediction on moving objects. Algorithms, 10(2):37, 2017.
QRCODE
 
 
 
 
 
                                                                                                                                                                                                                                                                                                                                                                                                               
第一頁 上一頁 下一頁 最後一頁 top