The objectives of this study were to develop a real-time tillage depth measurement system for agricultural tractor performance analysis and then to validate these configured systems through soil non-penetration tests and field experiment during plow tillage. The real-time tillage depth measurement system was developed by using a sensor fusion method, consisting of a linear potentiometer, inclinometer, and optical distance sensor to measure the vertical penetration depth of the attached implement. In addition, a draft force measurement system was developed using six-component load cells, and an accuracy of 98.9% was verified through a static load test. As a result of the soil non-penetration tests, it was confirmed that sensor fusion type A, consisting of a linear potentiometer and inclinometer, was 6.34–11.76% more accurate than sensor fusion type B, consisting of an optical distance sensor and inclinometer. Therefore, sensor fusion type A was used during field testing as it was found to be more suitable for use in severe working environments. To verify the accuracy of the real-time tillage depth measurement system, a linear regression analysis was performed between the measured draft and the predicted values calculated using the American Society of Agricultural and Biological Engineers (ASABE) standards-based equation. Experimental data such as traveling speed and draft force showed that it was significantly affected by tillage depth, and the coefficient of determination value at M3–Low was 0.847, which is relatively higher than M3–High. In addition, the regression analysis of the integrated data showed an R-square value of 0.715, which is an improvement compared to the accuracy of the ASABE standard prediction formula. In conclusion, the effect of tillage depth on draft force of agricultural tractors during plow tillage was analyzed by the simultaneous operation of the proposed real-time tillage depth measurement system and draft force measurement system. In addition, system accuracy is higher than the predicted accuracy of $\pm$ 40% based on the ASABE standard equation, which is considered to be useful for various agricultural machinery research fields. In future studies, real-time tillage depth measurement systems can be used in tractor power train design and to ensure component reliability, in accordance with agricultural working conditions, by predicting draft force and axle loads depending on the tillage depth during tillage operations.

This study aims to design, develop, and evaluate the traction performance of an electric all-wheel-drive (AWD) tractor based on the power transmission and electric systems. The power transmission system includes the electric motor, helical gear reducer, planetary gear reducer, and tires. The electric system consists of a battery pack and charging system. An engine-generator and charger are installed to supply electric energy in emergency situations. The load measurement system consists of analog (current) and digital (battery voltage and rotational speed of the electric motor) components using a controller area network (CAN) bus. A traction test of the electric AWD tractor was performed towing a test vehicle. The output torques of the tractor motors during the traction test were calculated using the current and torque curves provided by the motor manufacturer. The agricultural work performance is verified by comparing the torque and rpm (T–N) curve of the motor with the reduction ratio applied. The traction is calculated using torque and specifications of the wheel, and traction performance is evaluated using tractive efficiency (TE) and dynamic ratio (DR). The results suggest a direction for the improvement of the electric drive system in agricultural research by comparison with the conventional tractor through the analysis of the agricultural performance and traction performance of the electric AWD tractor.

Machine vision with deep learning is a promising type of automatic visual perception for detecting and segmenting an object effectively; however, the scarcity of labelled datasets in agricultural fields prevents the application of deep learning to agriculture. For this reason, this study proposes weakly supervised crop area segmentation (WSCAS) to identify the uncut crop area efficiently for path guidance. Weakly supervised learning has advantage for training models because it entails less laborious annotation. The proposed method trains the classification model using area-specific images so that the target area can be segmented from the input image based on implicitly learned localization. This way makes the model implementation easy even with a small data scale. The performance of the proposed method was evaluated using recorded video frames that were then compared with previous deep-learning-based segmentation methods. The results showed that the proposed method can be conducted with the lowest inference time and that the crop area can be localized with an intersection over union of approximately 0.94. Additionally, the uncut crop edge could be detected for practical use based on the segmentation results with post-image processing such as with a Canny edge detector and Hough transformation. The proposed method showed the significant ability of using automatic perception in agricultural navigation to infer the crop area with real-time level speed and have localization comparable to existing semantic segmentation methods. It is expected that our method will be used as essential tool for the automatic path guidance system of a combine harvester.