Intelligent Surface Recognition for Autonomous Tractors Using Ensemble Learning with BNO055 IMU Sensor Data DOI Creative Commons

Phummarin Thavitchasri,

Dechrit Maneetham,

Padma Nyoman Crisnapati

и другие.

Agriculture, Год журнала: 2024, Номер 14(9), С. 1557 - 1557

Опубликована: Сен. 9, 2024

This study aims to enhance the navigation capabilities of autonomous tractors by predicting surface type they are traversing using data collected from BNO055 Inertial Measurement Units (IMU sensors). IMU sensor were a small mobile robot driven over seven different floor surfaces within university environment, including tile, carpet, grass, gravel, asphalt, concrete, and sand. Several machine learning models, Logistic Regression, K-Neighbors, SVC, Decision Tree, Random Forest, Gradient Boosting, AdaBoost, XGBoost, trained evaluated predict based on data. The results indicate that Forest XGBoost achieved highest accuracy, with scores 98.5% 98.7% in K-Fold Cross-Validation, respectively, 98.8% 98.6% an 80/20 State split. These findings demonstrate ensemble methods highly effective for this classification task. Accurately identifying types can prevent operational errors improve overall efficiency systems. Integrating these models into tractor systems significantly adaptability reliability across various terrains, ensuring safer more efficient operations.

Язык: Английский

Applications of Key Autonomous Navigation Technologies for Unmanned Agricultural Tractors: A Review DOI Open Access
Jiwei Qu, Zhe Zhang,

Zheyu Qin

и другие.

Опубликована: Фев. 7, 2024

Unmanned agricultural tractor (UAT) represents the advanced stage of autonomous navigation and serves as a core technology in production. It reduces operator’s workload, improves operational accuracy efficiency. This article reviews three aspects key technologies for UATs: perception, path planning tracking, motion control. The advantages, shortages these on UATs are clarified by analyzing technical principles current research status. We conducted summaries analyses existing unmanned solutions different application scenarios order to identify bottleneck issues. Based analysis applicability UATs, it can be seen that fruitful progresses have been achieved. review also summarizes common problems technologies. sharing integrating multi-source data relatively weak. There is an urgent need high-precision, high-stability sensing equipment. universality methods, efficiency precision tracking improved. necessary develop high reliability electrical control modules enhance performance. Overall, sensors, high-performance intelligent algorithms, reliable hardware factors promoting development UAT technology.

Язык: Английский

Процитировано

6

Path Tracking Control of a Large Rear-Wheel–Steered Combine Harvester Using Feedforward PID and Look-Ahead Ackermann Algorithms DOI Creative Commons

Shaocen Zhang,

Qingshan Liu,

Haihui Xu

и другие.

Agriculture, Год журнала: 2025, Номер 15(7), С. 676 - 676

Опубликована: Март 22, 2025

Autonomous driving solutions for agricultural machinery have advanced rapidly; however, large-wheeled harvesters present unique challenges compared to traditional vehicles. Specifically, the 5.4 m cutting width, 9.2 minimum turning diameter, and rear-wheel–steered configuration demand specialized path tracking steering methods. To address these challenges, this study developed an integrated system combining feedforward PID Look-Ahead Ackermann (LAA) algorithms with sensors, actuators, embedded control platform. Field experiments indicated that maintained average lateral deviation of approximately 5 cm on straight-line paths, slightly larger errors observed only during or alignment maneuvers. Additionally, a “three-cut” method was implemented, which enhanced accuracy prevented crop damage at headland turns. Successful field tests confirmed robustness system, highlighting its practical potential production-level autonomous harvesting.

Язык: Английский

Процитировано

0

Sensing and Perception in Robotic Weeding: Innovations and Limitations for Digital Agriculture DOI Creative Commons
Redmond R. Shamshiri,

Abdullah Kaviani Rad,

Maryam Behjati

и другие.

Sensors, Год журнала: 2024, Номер 24(20), С. 6743 - 6743

Опубликована: Окт. 20, 2024

The challenges and drawbacks of manual weeding herbicide usage, such as inefficiency, high costs, time-consuming tasks, environmental pollution, have led to a shift in the agricultural industry toward digital agriculture. utilization advanced robotic technologies process serves prominent symbolic proof innovations under umbrella Typically, consists three primary phases: sensing, thinking, acting. Among these stages, sensing has considerable significance, which resulted development sophisticated technology. present study specifically examines variety image-based systems, RGB, NIR, spectral, thermal cameras. Furthermore, it discusses non-imaging including lasers, seed mapping, LIDAR, ToF, ultrasonic systems. Regarding benefits, we can highlight reduced expenses zero water soil pollution. As for obstacles, point out significant initial investment, limited precision, unfavorable circumstances, well scarcity professionals subject knowledge. This intends address advantages associated with each technologies. Moreover, technical remarks solutions explored this investigation provide straightforward framework future studies by both scholars administrators context weeding.

Язык: Английский

Процитировано

2

Modelling the Dynamics of a Wheeled Mobile Robot System Using a Hybridisation of Fuzzy Linear Quadratic Gaussian DOI Open Access

Sairoel Amertet Finecomess,

Girma Gebresenbet,

Hassan M. Alwan

и другие.

Опубликована: Июнь 21, 2024

Wheeled mobile robot dynamics and suitable controller design are challenging but rewarding fields of study. By understanding the wheeled robots, it could be possible to hybrid control schemes for robots. Since involve combining individual methods create a more effective overall strategy. This can done in variety ways, such as using fuzzy linear quadratic Gaussian control. We were therefore inspired dynamic models their designs by examining robot. The novelty current paper is hybridize different robots order get better performance. Entire systems simulated MATLAB/SIMULINK environment. results obtained settling time response FLQG 87.1% over LQG; this study compared effectiveness previous controllers external disturbances found peak amplitude improvements 71.25%. Therefore, proposed use with

Язык: Английский

Процитировано

1

A Novel Fuzzy Logic Switched MPC for Efficient Path Tracking of Articulated Steering Vehicles DOI Creative Commons
Xuanwei Chen, Jiaqi Cheng, Huosheng Hu

и другие.

Robotics, Год журнала: 2024, Номер 13(9), С. 134 - 134

Опубликована: Сен. 5, 2024

This paper introduces a novel fuzzy logic switched model predictive control (MPC) algorithm for articulated steering vehicles, addressing significant path tracking challenges due to varying road conditions and vehicle speeds. Traditional single-model parameter-based controllers struggle with errors computational inefficiencies under diverse operational conditions. Therefore, kinematics-based MPC is first developed, showing strong real-time performance but encountering accuracy issues on low-adhesion surfaces at high Then, 4-DOF dynamics-based designed enhance stability. The proposed solution strategy, integrating system that dynamically switches between algorithms based error, time, heading angle indicators. Subsequently, simulation tests are conducted using SIMULINK ADAMS verify the of algorithm. results confirm this fuzzy-based can effectively mitigate drawbacks approaches, ensuring precise, stable, efficient across adhesion

Язык: Английский

Процитировано

1

Intelligent Surface Recognition for Autonomous Tractors Using Ensemble Learning with BNO055 IMU Sensor Data DOI Creative Commons

Phummarin Thavitchasri,

Dechrit Maneetham,

Padma Nyoman Crisnapati

и другие.

Agriculture, Год журнала: 2024, Номер 14(9), С. 1557 - 1557

Опубликована: Сен. 9, 2024

This study aims to enhance the navigation capabilities of autonomous tractors by predicting surface type they are traversing using data collected from BNO055 Inertial Measurement Units (IMU sensors). IMU sensor were a small mobile robot driven over seven different floor surfaces within university environment, including tile, carpet, grass, gravel, asphalt, concrete, and sand. Several machine learning models, Logistic Regression, K-Neighbors, SVC, Decision Tree, Random Forest, Gradient Boosting, AdaBoost, XGBoost, trained evaluated predict based on data. The results indicate that Forest XGBoost achieved highest accuracy, with scores 98.5% 98.7% in K-Fold Cross-Validation, respectively, 98.8% 98.6% an 80/20 State split. These findings demonstrate ensemble methods highly effective for this classification task. Accurately identifying types can prevent operational errors improve overall efficiency systems. Integrating these models into tractor systems significantly adaptability reliability across various terrains, ensuring safer more efficient operations.

Язык: Английский

Процитировано

1