Title

Motion Planning and Control Strategies for Automated and Remote Driving Systems

Alternative Title

자율주행 및 원격주행 시스템을 위한 모션 계획 및 제어 전략

DGIST Authors

Ji Hwan Seo ; Kyoung-Dae Kim ; Yongsoon Eun

Advisor

김경대

Co-Advisor(s)

Yongsoon Eun

Issued Date

2025

Awarded Date

2025-08-01

Type

Thesis

Description

Automated driving, remote driving, motion planning, driving safety, ride comfort

Table Of Contents

Contents
Abstract i
List of Contents iv
List of Tables vii
List of Figures viii
List of Algorithms xii

1 Introduction 1
1.1 Background and Motivation 1
1.2 Contributions and Outline of Dissertation 4

2 Related Works 11
2.1 Model Predictive Control based Motion Planning 11
2.2 Dynamic Look-ahead Distance for Pure Pursuit 14
2.3 Prediction-based Methods to Mitigate Delay Impacts in Remote Driving Systems 17

3 A Discrete-time Linear Model Predictive Control for Motion Planning with Adaptive Cruise Control and Overtaking 25
3.1 Overall Structure 25
3.2 Vehicle Modeling 27
3.2.1 Kinematic Vehicle Model 27
3.2.2 Linearization 29
3.2.3 Discrete-time Linearized Vehicle Model 30
3.3 System Constraints 31
3.3.1 Representation of Lateral Displacement for Collision Avoidance 31
3.3.2 Lateral Deviation Constraints 32
3.3.3 Motion Limits 35
3.4 Target Speed Calculation 35
3.4.1 Non-sideslip Condition 36
3.4.2 Vehicle Following 37
3.4.3 Target Speed Determination Logic 38
3.5 Cost Function 39
3.5.1 Cost Function Design 39
3.5.2 Variable Weights for Smooth Lateral Motion 41
3.5.3 Lateral Error Cost Weight for Smooth Overtaking 42
3.6 Evaluation 44
3.6.1 Simulation Setup 44
3.6.2 Path Tracking Performance 47
3.6.3 Ride Comfort Quality 50
3.6.4 Collision Avoidance 53

4 A Systematic Look-ahead Distance Tuning Method for Pure Pursuit to Improve Path Tracking Performance 60
4.1 Preliminaries 60
4.1.1 Pure Pursuit 60
4.1.2 Tracking Error Model 62
4.1.3 Eigenvalue Analysis for Linear System 63
4.1.4 A Brief Review of Previous Work 65
4.1.5 Limitations of the Previous Work 66
4.2 Refined Look-ahead Distance Design Process 67
4.2.1 Tracking Error Model with Precisely Approximated Pure Pursuit 67
4.2.2 Look-ahead Distance Design Using Eigenvalue Analysis 70
4.2.3 Minimum Look-ahead Distance 71
4.3 Evaluation 72
4.3.1 Simulation Setup 72
4.3.2 Path Tracking Performance 74

5 Control Strategy for Improving Remote Driving Ability under Consecutive Packet Loss using Adaptive Hybrid Time Series Forecaster 77
5.1 Basic Idea 77
5.2 Proposed Strategy for Mitigating the Impact of Consecutive Packet Loss 78
5.3 Preliminary: MES-LSTM 81
5.3.1 MES-LSTM 81
5.3.2 Limitations of MES-LSTM 84
5.4 Adaptive MES-RNN 86
5.4.1 Smoothing Coefficient Updater 87
5.4.2 Dataset Generation for SC-Updater Training 90
5.4.3 Loss Function for SC-Updater Training 91
5.4.4 Preparation of Trend Predictor Training 93
5.4.5 Other Enhancements 94
5.5 Adaptive Intervention Time Threshold Design 94
5.5.1 Function Design 95
5.5.2 Parameter Fitting Process 98
5.5.3 Threshold Determination Logic 100
5.6 Evaluation 100
5.6.1 Experimental Setup for Prediction Model Test 101
5.6.2 Verification of ES Model Stability Condition 107
5.6.3 Prediction Accuracy Comparison 109
5.6.4 Computational Efficiency of AMES-RNN 114
5.6.5 Simulation Setup for Remote Driving Ability Test 115
5.6.6 Simulation Results 119

6 Conclusions of Dissertation 125

References 130
A Detailed Derivation Process 152
A.1 Approximation of Pure Pursuit Control Law around Circular Path 152

URI

https://scholar.dgist.ac.kr/handle/20.500.11750/59793
http://dgist.dcollection.net/common/orgView/200000893573

DOI

10.22677/THESIS.200000893573

Degree

Doctor

Department

Department of Electrical Engineering and Computer Science

Publisher

DGIST

Show Full Item Record