|
An Accurate Path Tracking Algorithm for Autonomous Vehicles Based on Pure Pursuit with Systematically Designed Look-ahead Distance and Sideslip Compensation |
|
Ji Hwan Seo, Sung Hoon Youn, Jungeun Kim, Kyoung-Dae Kim(DGIST, Korea) |
| Path tracking performance of Pure Pursuit (PP) is degraded by using an improper Look-ahead Distance (LAD) and ignoring the effects of dynamics. To address such issues, we propose a PP-based path tracker that quickly reduces tracking errors and considers the sideslip by dynamics. For fast tracking error reduction, a dynamic LAD is designed by mathematically analyzing the change in tracking errors according to the LAD. Also, to compensate for the sideslip effect, the desired steering angle is calculated using the sideslip integrated-PP geometry. As a result, the proposed method showed improved tracking performance compared to classical PP as well as existing improved PP. |
|
|
FS-ACO: An Algorithm for Unsafe U-Turn Detours in Service Vehicle Route Optimization Applications |
|
Tyler Parsons, Jaho Seo(Ontario Tech University, Canada) |
| In this study, an algorithm is proposed to yield the shortest distance detours to resolve unsafe u-turns in routing applications. In service routing operations such as waste collection and snowplowing, it is unsafe for these large vehicles to perform u-turns at intersections. Rather than doing a u-turn, the operator must find the shortest detour that leads back to the intersection to continue in the desired travel direction. To solve this problem, a forward-searching ant colony optimization (FS-ACO) algorithm is proposed. Experimental results show that the proposed algorithm can yield optimal detours for a collection of u-turn examples. |
|
|
Deploying multiple vehicles for snow plowing using Smart Selective Navigator and its effect |
|
Farhad Baghyari, Jaho Seo(Ontario Tech University, Canada) |
| Winter operations, such as snowplowing and salt spreading, have far-reaching implications for businesses, road safety, and mobility. This paper investigates the effect of deploying multiple vehicles for the same set of streets in winter operations using the Smart Selective Navigator (SSN) method. The study focuses on selected regions of the City of Oshawa. A mathematical model and problem formulation are presented, outlining the representation of the road network and variables used in the SSN method. The findings highlight the potential benefits of deploying multiple vehicles in winter operations, but careful consideration of trade-offs is necessary. |
|
|
Robust path following control of autonomous surface vessels |
|
Quoc Van Tran, Vinh Van Nguyen, Hoang Quang Nguyen(Hanoi University of Science and Technology, Viet Nam) |
| This work presents a robust path following control method for autonomous surface vessels (ASVs) based on
sliding mode control. To describe the control problem, the desired path of the vehicle is parameterized by one parameter and a parallel-transport path frame. The error dynamics of the path following system consisting of the position following error and the parameter speed’s error are subsequently derived in the body-fixed coordinates of the vessel. We develop a robust force controller to steer the vehicle’ velocity to track the reference velocity asymptotically and exponentially fast. Rigorous
stability analysis is provided to show the effectiveness of the proposed control method. |
|
|
SRN-CRL: Sparse Reward Navigation via Curiosity-driven Reinforcement Learning |
|
Ji Sue Lee, Jun Moon(Hanyang University, Korea) |
| In this paper, we propose a reinforcement learning (RL) based approach for navigation of mobile robots under sparse reward via curiosity-driven exploration strategy. It is well known that RL shows remarkable performance in practical situations, most real-world environments cannot be given as a dense reward, in which case the RL agent may not achieve the optimal action correctly, leading to unexpected navigation in mobile robots. To address this limitation, we design the sparse reward navigation via curiosity-driven reinforcement learning. |
|
|
Frenet Frame based Local Motion Planning in Racing Environment |
|
Min Seong Kim, jeon hyeok Lee, Taek Lim Kim, Tae-Hyoung Park(Chungbuk National University, Korea) |
| existing trajectory planning studies have demonstrated good performance in environments with static objects or when avoiding racing vehicles in straight-line sections, but accidents occur in environments with large numbers of vehicles and high curvature. To address these issues, this paper proposes an algorithm that ensures stability even in complex environments by considering the uncertainty of motion prediction for multiple vehicles. By considering the uncertainty of vehicle motion, we design the cost and generate an optimized path based on sampling. The CarMaker simulator is utilized to verify the proposed algorithm. |
|