A novel embedded system for tractor implement performance mapping

Abstract

The agricultural tractor serves as a pivotal power source on the farm. However, the optimal utilization of tractor power has often been hindered by factors such as non-matching implement sizes and suboptimal operating skills, leading to underutilization of power potential across diverse field conditions. In response, a study was initiated to formulate a cost-effective embedded system capable of on-the-go digital display and recording of various field performance parameters, ensuring precision in geographical location tracking. An embedded system, seamlessly incorporating a global positioning system (GPS), was successfully integrated onto an agricultural tractor to facilitate the mapping of tractor-implement performance. The developed embedded system employed Hall effect sensors for measuring theoretical speeds at both tractor rear drive wheels, as well as the actual speed at the tractor front wheel. Additionally, forces at the tractor’s three-point linkages were quantified with the assistance of a three-point hitch dynamometer, and tillage depth at the tractor rockshaft was measured using a rotary potentiometer. The obtained data exhibited exemplary results. This comprehensive and integrated system demonstrated the capability to measure, display and record real-time performance parameters, coupled with their precise geographical locations. The system underwent extensive and successful field demonstrations, particularly in the spatial mapping of tractor-implement field performance, utilizing a mounted offset disk harrow and a rigid tine cultivator.

Keywords:

• Tractor implement performance parameters
• GPS
• transducer
• spatial maps
• precision agriculture

REVIEWING EDITOR:

• D T Pham, University of Birmingham, United Kingdom

SUBJECTS:

• Agricultural Engineering
• Agricultural Development
• Agriculture & Environmental Sciences

1. Introduction

Precision farming has predominantly concentrated on mitigating chemical and fertilizer expenses and optimizing crop yield through timely harvesting, often overlooking additional contributors to overall farm production costs (Kushwah et al., 2023; Parashunath et al., 2017). Tractors have emerged as pivotal power units for land preparation, crop maintenance operations and transportation within the crop production industry in India and various other countries. The majority, approximately 85%, of Indian farmers fall under the small and marginal categories (Anonymous, 2018), operating under challenging socioeconomic conditions. A significant proportion of farmers engage contractors for field operations without discerning the quality of service, and contractors, in turn, may not optimize the tractor-implement combinations, leading to inefficient field operations, heightened labor and fuel consumption, and additional expenses for farmers due to overcharges and suboptimal work quality. Land preparation and field maintenance costs alone account for around 20–25% of the total production cost (Abdul Wahid et al., 1993). Enhanced profitability in crop production necessitates more efficient utilization of machinery resources during field operations. Addressing the operational efficiency of tractor-implement systems has been a subject of intensive research, with a focus on maximizing work output or minimizing fuel consumption through proper matching and optimization of tractor-implement operations (Nataraj et al., 2021). Rising costs of fuel, labor, repair and maintenance underscore the importance of efficient tractor-implement operations. Studies have indicated potential fuel consumption savings of up to 20% with techniques, such as Gear-Up Throttle-Down (M. Schrock et al., 1986; Grogan et al., 1987).

Advancements in computer and electronics technology, particularly microcontrollers and microprocessors, have facilitated measurements of tractor-implement field performance, such as wheel torque, forces acting on tractor wheels, tillage depth, PTO torque, pitch and roll angles and fuel flow rate (A Alsuhaibani et al., 2006; Yahya et al., 2009). Embedded systems, due to their enhanced processing speed, durability, compactness, and large data storage capacity, have become instrumental in acquiring, scanning, monitoring, and processing data (Bhagat et al., 2023; Chowdhury et al., 2023; Chouriya et al., 2023). Strain gauges were incorporated at vertical smooth blade to measure the soil resistance in real time with the use of global positioning system (GPS; I. Adamchuk et al., 2001). Some researchers also engineered draft force sensing devices with the aid of load cells (A. A. Kumar et al., 2016; Upadhyay & Raheman, 2020). The evolution has led to the development of on-board tractor performance monitoring systems, and the integration of GPSs with performance monitors has given rise to precision technology, creating a demand for comprehensive systems (A. A. Kumar, Tewari, Gupta, & Kumar, 2017; Scarlett, 2001; Van Bergeijk & Goense, 2015; Yule et al., 1999). Such integrated systems offer managers, engineers and farmers a deeper understanding, enabling precise control of input applications at specific field locations to enhance crop yield while minimizing environmental impacts. For instance, identifying areas with high tractor slip on a map can guide remedial actions to minimize input costs.

This study introduces an embedded system integrated into a two-wheel-drive tractor, TAFE SAMRAT 4410, recording tractor implement performance parameters continuously along its path. The collected data enables the creation of spatial maps for various parameters, offering farmers visual insights (in the form of colors) into the performance of operations and providing a solution to assess the quality of work received.

2. Materials and methods

An embedded system was devised for the comprehensive monitoring of tractor-implement performance. The system encompasses essential components, such as a GPS module and transducers, designed to monitor critical parameters including forward speed, rear-wheel speed, depth of operation, slip and draft (Figure 1). The specifications pertaining to the diverse transducers employed within the instrumentation package have been meticulously detailed and are presented in Table 1.

Figure 1. (A) GPS module; (B) Hall effect sensor; (C) Rotary potentiometer; (D) Three-point hitch dynamometer.

Table 1. Transducers and their specifications.

S. No	Transducer	Specifications
1.	Global positioning system U-blox Neo 7 M	Horizontal position accuracy 2.5 m, Maximum navigation update rate 10 Hz, Velocity accuracy 0.1 m/s
2.	Load cell RC02C Monad Ltd.	Type compression/tension, capacity 2000 kg, Supply voltage 10 V
3.	Hall effect sensor 44E	Latch type, Operating voltage 4.2–24 V
4.	Rotary potentiometer	No. of turns 1, Resistance 10kΩ
5.	Microcontroller	ATmega 2560, Flash memory 256 KB, Clock speed 16 MHz

2.1. Tractor global positioning system (GPS)

The determination of the latitudinal and longitudinal coordinates of specific points within the field was facilitated through the utilization of a U-blox GPS module (Neo-7 series), which was connected to the microcontroller. This module, recognized for its adaptability and cost-effectiveness, presented a compact 16 × 12.2 × 2.4 mm framework (Figure 1(A)). The NEO-7 modules’ diminutive architecture, in conjunction with its power and memory configurations, renders them particularly well-suited for deployment in battery-operated mobile devices constrained by stringent cost and space considerations. The National Marine Electronics Association (NMEA) signals at a frequency of 10 Hz received by the GPS module, featuring a ceramic antenna for enhanced communication capabilities. Real-time recording of the geo-position of the tractor-implement in relation to the aforementioned performance parameters was effectuated in the field of IIT Kharagpur (Figure 2). The acquired data underwent post-processing utilizing ArcGIS 10.0, a Geographic Information System (GIS) software developed by ESRI, thereby generating tractor-implement performance spatial maps. This post-processing activity aimed to facilitate site-specific analyses of variations in tractor-implement field performance.

Figure 2. Geo position data points on the field.

2.2. Transducer

Hall effect sensors were employed to measure both the actual and theoretical speeds of the tractor as it traversed the field (Figure 1(B)). Specifically, one Hall sensor was affixed to the front wheel of the tractor to ascertain its actual speed, while two Hall sensors were strategically positioned on each rear wheel to gauge the theoretical speed of the tractor. Small magnetic components were affixed equidistantly to the wheel rim to facilitate the measurement of revolutions per minute (RPM) as shown in Figure 3(A). The passage of the Hall sensor through a magnetic field resulted in the generation of a voltage signal. Over a single revolution, the sensor produced a count of pulses contingent upon the number of magnets affixed to the wheel rim. Subsequently, the angular speeds were multiplicatively adjusted by their respective rolling radii, estimated by measuring the distance traveled in one revolution divided by 2*π within the microcontroller, culminating in the computation of both the actual speed (V_a) and theoretical speed (V_t) of the tractor (Kumar, Tewari, Gupta, & Pareek, 2017; Mahore et al., 2022). These speeds were used in the computation of slip (Eq. (1)(1) $$\mathrm{\text{Slip}},\mathrm{\%}=\left(1-\frac{V_a}{V_t}\right)\times 100$$(1) ). (1) $$\mathrm{\text{Slip}},\mathrm{\%}=\left(1-\frac{V_a}{V_t}\right)\times 100$$(1)

Figure 3. Setup for (A) speed measurement; (B) depth measurement.

A rotary potentiometer of single rotation with a resistance of 10 kΩ was installed on the tractor rockshaft to measure the tillage depth of various implements within actual field conditions (Figure 1(C)). The device was affixed to an L-shaped bracket utilizing a jam nut, and its knob was integrated into a bush linked to the iron bar on the rockshaft of the hydraulic system (Figure 3(B)). The linear motion of the lift arm in the vertical plane was conveyed to the potentiometer, inducing both clockwise and counter-clockwise rotations. Consequently, variations in the position of the lift arm correspondingly altered the output voltage of the rotary potentiometer (Gupta et al., 2019). The output wires of the potentiometer were linked to the identical microcontroller employed for wheel slip measurement (Arduino Mega 2560).

Draft measurements for various mounted implements were conducted utilizing a three-point linkage dynamometer (Alimardani et al., 2008). This dynamometer, characterized as a universal three-point linkage variant compatible with any tractor-implement combination, featured a detachable frame constructed as a built-in sensor. The front end of the frame was affixed to the tractor, while the rear end was attached to the implement. The dynamometer incorporated three load cells distributed within three distinct sensing units to detect and measure the draft of the implement (Figure 1(D)). The comparatively low output from these load cells, an INA125P IC was utilized to amplify the output signal. The data acquired from all transducers and the GPS module were recorded on the computer using PuTTY software and subsequently retrieved in Microsoft Excel for further analysis. The selection of these transducers was grounded in specific criteria, including cost-effectiveness, low power consumption, ready availability, high linearity, and robust dynamic response. These considerations were paramount in guiding the choice of transducers for the study.

3. Calibration of transducer

The static calibration of all installed transducers was conducted in a laboratory setting prior to their application in the field. The transducers produced digital output voltages in response to applied forces. A standard procedure involving both ascending and descending load forces was employed to establish a linear relationship between the applied forces and their corresponding digital values, while also assessing hysteresis effects on the transducers. Across all calibration tests conducted, the obtained correlation coefficients consistently approached unity for the respective transducers.

For the compression force measurement setup (Figure 4(A)), a hydraulic jack was positioned on a weighing platform, with a load cell situated between the hydraulic jack and a fixed iron bar at the top. Force was applied to the load cell using the hydraulic jack, as one end of the cell remained fixed on the iron bar. The load was transferred to the weighing platform, and readings of the applied force were measured. Digital outputs from the load cell were recorded for every applied load, and a calibration curve correlating the compression force applied with the digital output was plotted (Figure 5(A)).

Figure 4. Calibration setup for (A) load cell in compression; (B) load cell in tension; (C) depth measurement.

Figure 5. Calibration curve (A) load cell in compression; (B) load cell 1 in tension; (C) load cell 2 in tension; (D) for depth measurement.

The tension force measurement setup (Figure 4(B)) featured a load crane for applying tension force, as displayed by a digital crane scale. The load cell was positioned between the crane scale and a fixed iron bar at the bottom. Two load cells were calibrated for tension forces corresponding to the two lower links. Calibration curves illustrating the relationship between the tension force applied and digital output were plotted (Figure 5(B,C)).

The tillage depth measurement setup (Figure 4(C)) involved placing the tractor along with the implement on a flat surface, denoted as the reference point and recording the digital output of the rotary potentiometer for this lift arm position. Subsequently, the tractor and implement were elevated using a hydraulic jack, the implement was lowered and its displacement was measured with a steel ruler. Changes in digital value corresponding to displacement were recorded, and the calibration curve for the potentiometer used to measure implement depth was depicted (Figure 5(D)).

4. Performance of developed embedded system

The schematic circuit diagram of the developed embedded system as shown in Figure 6, served as the blueprint for the construction of the embedded system (Figure 7). The resultant embedded system, designed as a plug-and-play device, featured all ports conveniently positioned on its outer surface. During system application, the transducers were effortlessly connected directly to the system, rendering it ready for use. The power supply for the developed system was sourced from a laptop. The output data, encompassing latitude, longitude, slip and draft, were prominently displayed on the LCD screen. Field experimentation was conducted utilizing a TAFE SAMRAT 4410, a 2WD tractor with a maximum power take-off (PTO) power of 33.8 kW. The designated test plot, characterized as sandy clay loam soil, exhibited a water content ranging from 104 to 121 g/kg (dry basis) and a dry bulk density of 1.59–1.67 Mg/m³ for the 15 cm topsoil layer. The tractor implements performance parameters were evaluated in field tests employing a rigid tine cultivator (9 tines and 2.16 m total width) and an offset-type disk harrow (14 discs and 1.78 m total width) as shown in Figure 8. Prior to each test, the soil cone index was measured with cone penetrometer. For each test, the measured draft was compared against the ASABE draft prediction (modified) protocol given by Eq. (2)(2) $$D_f=K_1\left\{V_j\left(A_1+B_1{V}_a+C_1{V}_a^2\right)\right\}{W}_I{d}_t+K_2$$(2) (ASAE, D497.4; Tiwari, 2006). Concurrently, validation tests were conducted for each transducer employed in the system. (2) $$D_f=K_1\left\{V_j\left(A_1+B_1{V}_a+C_1{V}_a^2\right)\right\}{W}_I{d}_t+K_2$$(2)

Figure 6. Circuit diagram of the developed embedded system.

Figure 7. Developed embedded system (A) front view; (B) inner view; (C) side view.

Figure 8. Field test with (A) cultivator; (B) harrow.

where D_f = Draft force of implement, N; V_j = soil texture parameter, constant; j = 1 for fine, 2 for medium, and 3 for coarse textured soil; d_t = tillage depth, cm; V_a = actual travel speed, km/h; W_I = width of implement, m; A, B and C = machine specific parameters; K₁ and K₂ = correction coefficients.

4.1. Validation of speed measurement

The output of the speed measurement, recorded through the implemented embedded digital system, underwent a comparative analysis against manually measured tractor speeds under tarmacadam conditions. The speed data obtained from both systems are presented in Figure 9, focusing on high gear ratios (H1, H2 and H3) and engine speeds of R1 = 1000 rpm, R2 = 1500 and R3 = 2000 rpm. Notably, the analysis revealed no discernible significant differences in any of the settings, as illustrated in Figure 9. The same results were observed by Gupta et al. (2019). A statistical examination of the data was further conducted utilizing a two-sample t-test, revealing no significant disparity (p value = 0.996) between the manually measured values and those obtained through the developed embedded system, with a significance level set at 5%.

Figure 9. Validation of speed measurement.

4.2. Validation of depth measurement

The depth measurements for both the cultivator and harrow were conducted using the implemented embedded system in the actual field. Post-test, manual depth measurements were performed at 15 randomly selected points in the field using a measuring scale. The comparison (Figure 10) indicated the absence of any significant differences. Furthermore, the average variation between the system-measured depth and manually measured depth for the cultivator was 6.74%, and for the harrow, it was 6.57%. These variations were well within the acceptable limit, affirming the satisfactory performance of the embedded system in depth measurement.

Figure 10. Validation of depth measurement.

4.3. Comparison of the observed draft with modified ASABE equation

In the field experiments, a rigid tine cultivator equipped with 9 tines and an offset harrow featuring 14 discs was utilized, with a dynamometer affixed to the tractor for both experiments. The draft measurements, obtained at equal time intervals for both the cultivator and harrow (Figure 11), revealed average draft values of 3522 N and 2973 N, respectively. Employing the measured speed and depth in the modified draft equation, predicted drafts for both implements were calculated, resulting in average values of 3693 N for the cultivator and 3141 N for the harrow. Total 56 observations were recorded for each test. The variations between the measured draft and the predicted draft were determined to be 4.63% for the cultivator and 5.34% for the harrow (Figure 12). Statistical analysis, employing a two-sample t-test (assuming equal variances), indicated no significant difference between the measured and predicted values for both the cultivator [−1.656 (110) = 0.100, p > 0.05] and the harrow [−1.857 (110) = 0.065, p > 0.05] at a 5% level of significance. This suggested a lack of substantial difference in the means of the measured and predicted values, and the high p value indicated that this difference is statistically no significant, providing compelling support for the acceptance of the null hypothesis.

Figure 11. Measured and predicted draft variations with time (A) cultivator; (B) harrow.

Figure 12. Comparison of measured and predicted draft for (A) cultivator; (B) harrow.

4.4. Spatial maps for performance parameters

The developed instrument system demonstrated its capability to measure, display and record real-time tractor performance parameters, including the travel speed of the tractor, wheel slippage, drawbar force and tillage depth, along with latitude and longitude during field operations. Spatial maps were generated for each parameter through post-processing of the acquired data using GIS software, specifically ArcGIS 10.0, for both the cultivator (Figure 13) and harrow (Figure 14). These spatial maps provided insights into the nature of the soil, field and highlighted the spatial variability of each parameter.

Figure 13. Spatial variability maps of (A) tillage depth; (B) draft force; (C) wheel slip; (D) travel speed for the cultivator.

Figure 14. Spatial variability maps of (A) tillage depth; (B) draft force; (C) wheel slip; (D) travel speed for disk harrow.

For instance, the spatial maps of tillage depth revealed negative values at the headland of the field for both the cultivator and disk harrow, indicative of implement lifting during turning. Similarly, draft maps illustrated the lowest draft at the field edges for both implements due to implement elevation. The spatial maps of slip exhibited negative values at the turning points for both implements, reflecting the skidding phenomenon of the front tires. Tractor speed maps depicted non-uniform speed patterns for the cultivator, particularly noticeable in barren land initially. Conversely, the speed map for the harrow showcased the uniform speed and lower draft in comparison to the cultivator, attributed to the smoothness and looseness of the soil surface. These spatial maps contributed to a comprehensive understanding of the field dynamics and the spatial variability of the measured parameters.

The field-tested system demonstrated commendable performance, with consistently accurate measurements affirming the efficacy of the embedded technology in operational contexts. The coefficient of variation values for all parameters fell within the acceptable range of 10% around the mean, aligning with anticipated variations under field conditions (Singh & Singh, 2011). The outcomes of these field trials underscore the tractor’s potential as a precision farming tool for spatial data collection. Additionally, the operator receives information that, when this smart system is put into place, will enable higher operating efficiency in Agriculture version 4.0 (Javaid et al., 2022). Identification and targeted remediation of areas characterized by high draft, uneven depth and excess wheel slip have led to significant savings in fuel, labor, and time resources.

5. Conclusion

The successful integration of an embedded system onboard a TAFE SAMRAT 4410 agricultural tractor facilitated the mapping of tractor-implement performance with geographical location. This comprehensive and integrated onboard instrumentation system exhibited the capability to measure, display and record real-time geo-positioning of the tractor-implement in the field concerning performance parameters, such as travel speed, drive wheel slippage, draft force and tillage depth. Static calibration tests conducted on the onboarded transducers demonstrated exceptional measurement linearity, with correlation coefficients approaching unity. The mapping system underwent rigorous field testing, proving its ability to function effectively in challenging and rugged field conditions. Validation of each system-measured parameter revealed a nonsignificant difference compared to manual measurements. The average variations observed for cultivator depth and draft were 6.74% and 4.63%, respectively, while for the disk harrow, they were 6.57% and 5.34%, respectively. Spatial maps depicting performance parameters provided valuable insights into system performance, field conditions and the quality of work conducted by operators. This system was found to possess considerable potential for the conduct of field testing and evaluation of emerging agricultural implements. Moreover, it demonstrated efficacy in establishing a comprehensive information database pertaining to the power and energy requisites of diverse tractor field operations. Additionally, the system proved instrumental in facilitating precision farming applications, specifically in the management of site-specific variability in tractor implement performance.

Authors’ contributions

AK and VKT: conceptualization, visualization and supervision. AK and AC: investigation and writing. AK, CG and PS: methodology. AC, MC and PB: validation. MC and PB: formal analysis. VKT, CG and PS: resources. AK and AC: data curation. AK, AC and VKT: writing, review and editing. All authors contributed to the article and approved the submitted version.

Acknowledgments

Authors would like to acknowledge Department of Agricultural and Food Engineering, IIT Kharagpur, for providing the necessary facilities for conducting the research work.

Disclosure statement

The authors affirm that they have no identifiable competing financial interests or personal relationships that might have influenced the work reported in this article.

Data availability statement

The original contributions presented in the study are included in the article; further inquiries can be directed to the corresponding author.

Additional information

Funding

The funding for this research was provided by the Agriculture and Food Engineering Department at Indian Institute of Technology Kharagpur (IIT Kharagpur), India (Institute ID: IR-O-U-0573).

Notes on contributors

Ajay Kushwah

Ajay Kushwah- Research Scholar

Arjun Chouriya

Arjun Chouriya- Research Scholar

V. K. Tewari

V. K. Tewari- Director, IIT Kharagpur, India

Chanchal Gupta

Chanchal Gupta- Scientist, CSIR-CMERI CoEFM, Ludhiana, India

Manojit Chowdhury

Manojit Chowdhury- Research Scholar

Prateek Shrivastava

Prateek Shrivastava- ICAR-NINFET, Kolkata, India

Preeti Bhagat

Preeti Bhagat- Research Scholar