상태 흐름 방법을 기반으로 실내 환경에서 외발 자전거 형 이동 로봇을 위한 경로 추종 플랫폼 개발

첸티안 ( Tean Chen ),유헌종 ( Heonjong Yoo ),이경환 ( Kyeonghwan Lee ) 한국농업기계학회 2020 한국농업기계학회 학술발표논문집 Vol.25 No.2

Localization is one of the important method for autonomous indoor robots to recognize it’s own position. In general, navigation of mobile robots is conducted using camera, Lider and GPS. But in case of indoor environment, GPS is unavailable. In this presentation, an autonomous indoor mobile robot, that is, a shuttle robot used a state flow method via ROS network: MATLAB and Linux high-level computers, IMU sensor, and then was able to obtain the cartesian coordinate information of the unicycle type mobile robot. After setting the pre-determined time based on the length of the path in the State Flow block, a path planning which is able to execute the work effectively is established using state flow algorithm. The state flow block produces time-series data sets which represents linear and angular velocities signals. Depending on the numerical values of the signal, the left and right motor rotational speed should be calculated through mobile robot forward kinematics. Several cases are considered: Case I) indicated the linear velocity is set positive certain value, and angular velocity is zero, so that the corresponding mobile robot moves forward. Case II) says that the linear velocity is set positive certain value and angular velocity is set positive certain value, which means the mobile robot turns right. Case III) says that the linear velocity is set positive certain value and angular velocity is set negative certain value, which means the corresponding mobile robot turns left. Case IV) says that the linear velocity is set negative certain value and angular velocity is set zero, which indicates the mobile robot moves backward. The effectiveness of the methods is demonstrated through desktop based developed indoor mobile robot’s control results.

Encoder-based localization method based on pre-built indoor environment map for 2-wheel agricultural indoor mobile robot

첸티안 ( Tean Chen ),조철현 ( Jo Chulhyun ),파블러 ( Pablo Vela ),유헌종 ( Heonjong Yoo ),이경환 ( Kyeonghwan Lee ) 한국농업기계학회 2021 한국농업기계학회 학술발표논문집 Vol.26 No.2

In an agricultural indoor environment like a greenhouse, it is hard to utilize GPS sensor for estimating mobile robot position. So odometry sensor i.g. encoder is needed for indoor agricultural robot localization. In a structured agricultural indoor environment, simultaneous localization and mapping method using multi-sensors such as Lidar, IMU, encoder to find a destination can be inefficient. we propose that encoder-based localization method based on a pre-built indoor environment map for efficient navigation of a 2-wheel agricultural indoor mobile robot. First, we constructed a 2D indoor map using Lidar and IMU data, and in the constructed map, for effective localization of the mobile robot, we divided the free space that the indoor mobile robot can navigate and the obstacles that interfere with the robot's moving in advance. And then to estimate the robot localization we adopt odometer increment model using encoder measurements of the indoor mobile robot. In order to evaluate the localization accuracy performance of our proposed method, we compared with the localization performance of the existing SLAM algorithms- LOAM, LeGo LOAM, A LOAM- for the position accuracy and odometry measurement accuracy with comparison to the position of the pre-built map. The position accuracy and odometry prediction accuracy of our proposed algorithm were evaluated as the average of 96.0% and 97.2% respectively. In the future, we will conduct further experiments in real agricultural indoor environments such as greenhouses.

음성 인식 기반 경로 추종 제어 플랫폼 개발

첸티안 ( Tean Chen ),조철현 ( Jo Chulhyun ),유헌종 ( Heonjong Yoo ),이경환 ( Kyeonghwan Lee ) 한국농업기계학회 2021 한국농업기계학회 학술발표논문집 Vol.26 No.1

In this presentation, the state flow method is utilized for path following indoor environment. Here, the novel voice recognition part is incorporated into path tracking problem using TCPIP communication. The two path following scenario was designed, and TCPIP receive block receives ASCII code from Android program. If the ASCII value is bigger than threshold value we set in the TCPIP receive block, the block itself chooses path 1, otherwise the block chooses path 2 in the real-time implementation. In order to implement the voice recognition based platform control, the Android program, MATLAB/SIMULINK program, Linux system are connected through ROS (Robot operating system) node connection. To summarize that, nodes are composed of three parts: the Android program, Linux system, MATLAB/SIMULINK program.

Optimal path navigation in the narrow indoor environment under the waypoint guidance of an autonomous agricultural shuttle vehicle

첸티안 ( Tean Chen ),조철현 ( Jo Chulhyun ),파블러 ( Pablo Vela ),유헌종 ( Heonjong Yoo ),이경환 ( Kyeonghwan Lee ) 한국농업기계학회 2022 한국농업기계학회 학술발표논문집 Vol.27 No.1

Accurate shuttle vehicle localization in the indoor environment on a precise map enables the mobile robot to estimate its position and orientation while moving in the indoor environment more efficiently. Trajectory tracking control is one of the fundamental techniques influencing a mobile robot's autonomous driving performance. MATLAB and Linux are high-level computers to equip the Inertial Measurement Unit (IMU), Velodyne VLP-16 channels LiDAR, and Encoder Sensors with the mobile robot platform. The mobile robot controls using a Robot Operating System (ROS) enabled robot, setup parameters for the differential wheels, and visualizes sensor data in a ROS robot visualization tool. In this paper, our model presents a new path following a method that integrates the pure pursuit algorithm and the state flow algorithm using the ROS Simulink model. The path following algorithm that performs autonomous waypoint navigation and obstacle avoidance method successfully localized and tested indoor environment. The accuracy improvement is demonstrated through several experimental results.

Active Path Planning Algorithm for Autonomous Mobile Robot Moving in Indoor Environment

첸티안 ( Tean Chen ),조철현 ( Chulhyun Cho ),파블러 ( Pablo Vela ),이경환 ( Kyeong-hwan Lee ) 한국농업기계학회 2022 한국농업기계학회 학술발표논문집 Vol.27 No.2

Accurate localization of mobile robots in indoor environment based on a precise map enables the mobile robots to estimate its position and orientation while moving in the indoor environment more efficiently. The objective of this study was to improve the mobile robot's navigation accuracy using the Rapidly Exploring Random Trees (RRTs) algorithm and the Hybrid A-Star algorithm, which can generate path planning capability in high-dimensional space for mobile robot localization. Localization and navigation for mobile robots were estimated using an inertial measurement unit, a LiDAR, and encoder sensors. The Robot Operating System (ROS) enabled the robot, set parameters for the differential wheels, and visualized sensor data in the visualization tool. The accuracy of the path planning method showed more than 91% success rate in mobile robot trajectory tracking. In the future, we will conduct experiments in more complicated areas, such as greenhouses and farms.

Fuzzy Logic Controller Design for an Agricultural Four-Wheel Independent Mobile Robot

파블로 ( Pablo Vela Ulloa ),첸티안 ( Chen Tean ),이경환 ( Kyeong-hwan Lee ) 한국농업기계학회 2021 한국농업기계학회 학술발표논문집 Vol.26 No.2

Due to their improved maneuverability in narrow spaces and increased stability, four-wheel independent mobile robots are becoming popular in several fields, such as agriculture, electrical vehicle, and planetary exploration. However, control algorithms is complicated owing to synchronization issues, mechanical constraints, and actuators equipped. This study presents a navigation controller based on fuzzy logic method for a low velocity autonomous agricultural vehicle built on a 4 wheel independent steering configuration. This paper explores the kinematic model of a 4 wheel independent steering robot and then real time 3D simulations using ROS and Gazebo. In the simulations, virtual GPS and IMU were used as sensors. Uneven terrains or obstacles were not considered for the simulated experiments. The simulation results show the capability of the fuzzy-logic controller in controlling a 4 wheel independent steering robot. In future, the simulated experiments will be compared with field experiments. Futhermore, the field experiments will be conducted in the condition of wheel slippage and low surface friction.

Caterpillar Equipped Mobile Robot Wokring on Rough Terrain based on Dynamic Window Approach

관티엔유엔 ( Tianyuan Guan ),첸티안 ( Chen Tean ),이경환 ( Kyeong-hwan Lee ) 한국농업기계학회 2020 한국농업기계학회 학술발표논문집 Vol.25 No.1

The robotic technologies are leading a technological revolution in agriculture. Although wheeled chassis is commonly used in most general landform, it does not have a great ability to deal in rough terrains. In this research, an autonomous caterpillar equipped robot was developed for outdoor agricultural tasks especially in rough terrains. The robot system consisted of vision system, localizing system, and high-level and low-level controller. The A-star searching and dynamic window approach were used as global planner and local planner respectively, which could generate a collision-free path and send final velocity commands to robot. The navigation simulation and actual navigation experiments was conducted. As results, the mobile robot could cruise certain path accurately with RMSE of lateral offset around 5 cm during whole navigation process. The accuracy can satisfy the demands of most agricultural tasks.

Lyapunov Controller for an Agricultural Four-Wheel Independent Mobile Robot

파블로 ( Pablo Vela Ulloa ),첸티안 ( Chen Tean ),이경환 ( Kyeong-hwan Lee ) 한국농업기계학회 2022 한국농업기계학회 학술발표논문집 Vol.27 No.2

Due to their improved maneuverability in narrow spaces and increased stability, four-wheel independent mobile robots are becoming popular in several fields, such as agriculture, electrical vehicle, and planetary exploration. However, control algorithms are complicated owing to synchronization issues, mechanical constraints, and actuators equipped. This paper explores the kinematic model of a 4 wheel independent steering robot and make use of a Lyapunov controller and A-star planner for navigate a low velocity autonomous agricultural vehicle built on a 4 wheel independent steering configuration. The experiments were conducted in an even surface and the robot was equipped with incremental encoders for the driving motors and absolute encoders for the steering motors as well as a GPS TDR-3000 and a IMU LPMS-IG1 for navigation. Uneven terrains or obstacles were not considered for the experiments. The results show the capability of the controller to navigate the 4 wheel independent steering robot and reach the desired goals. In future, a local planner will be considered for obstacle avoidance and complex field experiments will be conducted in conditions of wheel slippage and low surface friction.

Counting of Dense Onions using Improved YOLOv3 Model for Onion Picking Robot

관티엔유엔 ( Tianyuan Guan ),첸티안 ( Chen Tean ),이경환 ( Kyeong-hwan Lee ) 한국농업기계학회 2020 한국농업기계학회 학술발표논문집 Vol.25 No.1

The deep learning technology have been applied in different domain successfully, it also has great potential in various agricultural applications. Real-time crop detection can provide a lot of information to agricultural robot and achieving various tasks. However the onions are small and in dense distribution, traditional detection model cannot perform well in this situation. In this research, an improved YOLOv3 model was proposed for dense onions detection. Onion images were collected at the conditions of different illuminations, different onions distributions, and occluded samples. The YOLOv3 model structure was simplified by truncating redundancy branches and layers. Then, the generalized intersection over union(GIoU) was introduced into loss function and so weights could be updated even no intersection between prediction and ground truth. The imbalance of inliers and outliers was reduced by applying focal loss method. The test result show that the proposed model was faster and has better generalization ability than original YOLOv3 model. The counting accuracy was over 96% in different onions distribution and the detection speed was improved by 30% than original method.