1 Introduction

Agriculture saw a significant transformation in the next ten years as a result of the usage of robotics and artificial intelligence (AI). Additionally, researchers agreed that sensors and equipment have considerably more potential than what we are now employing [1,16,17,18], robotic grippers [7, 19]. Some systems used inserting needle/pin-type structure as a picking mechanism that inserts into the seedling soil media or plugs for extracting it [20,21,22,23,24,25,26,27,28,29,30]. Khadatkar et al. [11] developed an embedded system for an automatic vegetable transplanter in which a stepper and a DC motors were used to index the seedlings in the seedling tray. The stepper motor provides longitudinal motion to the metering shaft on which rotating fingers are mounted. The ground wheel provides the rotational motion to the fingers which helps to strike out the seedling from the portray. The shaft of the DC motor is used to rotate the feed roller so that the next seedling row comes to the striking place. Due to the striking action, the seedling gets ejected from the cavity of the nursery tray and jumps into the planting funnel of the transplanter. A machine vision system (MVS) or sensors were used to locate the seedling and empty cells on the nursery tray [18]. The system consisted of a sensor or a camera and computer for programming. The MVS was used to detect empty cells in high-density protray and guide the end-effector to move to the seedling location. A servomotor controlled jaw-type gripper was used to extract the seedling from the nursery tray used in a robotic tranplanter [7]. Choi et al. [22] developed a seedling pick–up device (gripper) to pick the seedling from tray and drop them at the pre-determined location from where the seedling gets planted. The device has pick–up pins that are actuated closed loop path to complete the pick and drop of the seedling. For the indoor vegetable cultivation Mao et al. [26] developed a pincette-type pick-up device for automatic transplanting of seedlings in greenhouse which can be retrofitted on a Gantry system of polyhouse. The pick-up dive has injection type needles (two numbers) to penetrate the root lump and grasp the seedling from the seedling tray. The grasped seedling transferred from the try and shifted laterally to for drop** in the destination pot. The manipulator uses Cartesian coordinate system, for movement of the end-effector to position in X and Y direction. Han et al. [28] developed a doorframe-typed swing mechanism for seedling extraction in automatic transplanter. The mechanism consisting of a path manipulator and two grippers to extract seedlings from the nursery tray and return quickly to the pick-up point for the next extraction by slowly moving the pins. Proper orientation of the seedling after release of seedling from pick-up device is very necessary. Moreover, mechanical systems sometimes proven better when equipped with electronic systems. Therefore, to meet the desired picking orientation after seedling release Yu et al. [30] developed a seedling picking device using planetary gear train with combined non-circular gear transmission. The gears were designed and meshed with each other in such a way that the extractor pick-up the seedling from cell and transferred to the planting position vertically.

The robotic transplanter uses grippers to hold the seedlings from the stems and then lift the seedling from the nursery tray by using a vision- or sensor-based system [7, 15, 18, 31,32,34,35,36,37,38,39]. Ting et al. [15] designed a robotic workcell for transplanting plugs from plug trays to growing flats. The main component of the workcell was the end-effector having “sliding-needles-with-sensor” gripper to extract, hold, and plant the plugs. The robot is only instructed to finish a transplanting cycle by the sensor after the gripper has securely grasped a plug. The cycle time was 2.54 s/plug. Ryu et al. [18] developed a robotic transplanter for bedding plant. The transplanter was made up of conveyors for plug trays, an end-effector, a manipulator, and a vision system. To position the end effector, the manipulator is equipped with two electrical linear motors. The end effector was pneumatically operated to grasp and hold the seedling till complete transplanting cycle. The success rates of transplanting were 97.8, 97.7 and 98.2% for 16-day cucumber seedlings, 13-day cucumber seedlings and 26-day tomato seedlings, respectively. Khadatkar et al. [7] report about a robotic transplanter (RT) consisting of a seedling pickup mechanism (SPM) and vehicle movement system (VMS). The SPM was used to extract the seedling from the nursery tray whereas, the VMS moves the robot forward using sensor for seedling detection. The developed RT takes 20 s to extract and transplant seedling successfully into the furrow.

Ishak et al. [32] developed a transplanter for the Gantry System operated under greenhouse conditions. The transplanter was automatically operated using a graphical user interface developed using Visual Basic 6.0 software. The transplanter has XZ-axis module, an auger, a gripper and a watering system. The XZ-axis module was operated with the help of two electric motors (a stepper and a DC motor). The average transplanting time of potted eggplant was 2 min 35 s.

Huang and Lee [34] developed a robotic arm based machine vision-guided system for gras** Phalaenopsis tissue culture plantlets. The system uses a gripper-gras** device suitable for gras** the PTCP plantlet. The grabbing point's three-dimensional coordinates were calculated using a binocular stereovision algorithm. The success rate of grasp and pick up the plantlet in the specified position was 78.2% when tested on 348 PTCP plantlets. Tong et al. [35] developed a machine vision based system to detect the state of seedlings in the nursery tray, locating unhealthy and empty seedling, guiding the manipulator to transplant the seedling. The system gives successful identification rate of 96.2% when tested on tomato seedlings. This technology could be used on automatic seedling transplanting robots. Hu et al. [37] developed a high-speed transplanting robot for plug seedling to improve the automation and efficiency for transplanting in greenhouse. It makes use of a pneumatic manipulator and a 2-DOF parallel translation mechanism. The mechanism to retrieve and place seedlings in a predetermined direction propelled the manipulator. The actual work cycle of single transplanting required 1.08 s.

Ndawula and Assal [38] developed a conceptual model of a 3DOF multi-gripper transplanting robot for pot seedlings and performed its kinematic analysis. According to simulation findings, the manipulator's accessible workspace contains the needed workspace, and a single transplanting cycle may produce six seedlings in 1.8 s. Rahul et al. [39] developed a 5R 2DOF parallel robot arm for handling paper pot seedlings in a vegetable transplanter. The gripper of the system performs pick and place operations at a desired path. The robot arm was tested and evaluated for picking and placing 15 numbers of actual paper pots of 80 g weight (5% mc) filled with vermicompost, soil and sand. With a maximum power usage of 20.47 W, the robot arm needed 2.1 to 2.4 s to pick and release a pot seedling from a distance of 116.6 mm.

Increased use of sensors and electronics has paved the application of robotics in many agricultural operations. Transplanting is one of the most labour demanding operations which needs to automate to reduce human drudgery involved in conventional practice as well as to perform the work in time. Again many attempts have been made to develop semi-automatic and automatic transplanters but still, farmers follow the traditional method i.e. manual. This may be due to the reason that either the technology is not as per their requirement or it is very expensive for them to purchase [8, 11]. Also, the availability of labour during peak season is the limiting factor to find for an alternative solution that require less human intervention. This paper aims to highlight the need for robotics in the development of a robotic transplanter for nursery-grown seedlings like chilli and tomatoes as well as to evaluate the system. More specifically the automatic seedling extractor (ASE) was developed to extract the seedling from the nursery portray using a jaw-type gripper automatically by gras** the seedling from the stem.

2 Materials and methods

2.1 Nursery seedlings

The physical properties of nursery seedlings of chilli (Variety: Pusa Jwala) and tomato (Variety: Abhilash) were used in this study. The criteria for seedlings selection were primarily based on degree of softness and stiffness. Specifically, chilli seedlings exhibit sturdiness and hard in nature while tomato seedlings were soft and tender when compared to other seedlings [40]. The seedling growing media of cocopeat, soil, perilite and vermiculite were used in ratio of 2:1:1:1, respectively. The 30 days old seedlings were used for testing the transplanting (Fig. 1). Seedling properties being essential parameter viz. weight, height, stem diameter, drop shatter height, critical canopy diameter, etc. needs to be measured while develo** the transplanter [7, 40].

Fig. 1
figure 1

Nursery seedlings of chilli and tomato grown in trays

Full size image

2.2 Design requirements

The mechanism used to extract the seedling from the tray should allow the manipulator to move freely and the gripper to secure the plant without snap** its stem. When develo** the robotic transplanter as shown below, the following design specifications should beconsidered.

  1. (1)

    The manipulator of ASE should move straight in the XY plane to extract the seedling before returning.

  2. (2)

    To securely grasp the seedling without breaking it, the gripper should open and close.

  3. (3)

    The seedling should be deposited directly into the delivery pipe while being released exactly above it.

  4. (4)

    To activate the directional control unit (DCU), the sensor must recognize the dropped seedling in the supply pipe.

  5. (5)

    The DCU should react and advance the entire robot.

2.3 Structure and components of RT

The RT consists of main frame, manipulator, seedling tray, gripper, programmable logic controller (PLC), Infrared (IR) sensor, directional control unit, motors, etc. (Fig. 2). Table 1 shows the overall specification of the developed transplanter. The seedling tray was kept on platform, the manipulator advances towards the nearest seedling kept on the platform, extract it and released into the delivery point. The IR sensor attached on the delivery pipe identifies the dropped seedling and then actuate the DCU to move the RT to next position. The movement of the manipulator and the gripper is synchronized in such a way that the gripper operates in the Z axis while the manipulator travels in the XY direction. The main components of RT are manipulator, gripper, seedling tray platform, planting device. The manipulator is used to move the extractor in XY plane and the gripper in Z plane. The manipulator and the gripper were shown in Fig. 3. The seedling tray platform was located just below the manipulator as it holds the seedling tray. The seedling tray was placed in such a way that it should not block the delivery point of the seedling during its release as shown in the Fig. 4. The platform has a hole of diameter 72 mm for the delivery of the seedling into the furrow. The planting device used here was a shoe-type furrow opener with compaction wheel for compacting the seedling in the furrow (Fig. 5).

Fig. 2
figure 2

Isometric view of robotic transplanter

Full size image
Table 1 Overall components of the developed transplanter
Full size table
Fig. 3
figure 3

Manipulator and the end-effector (jaw-type gripper)

Full size image
Fig. 4
figure 4

Seedling tray platform with seedling tray (all dimensions in mm)

Full size image
Fig. 5
figure 5

Shoe-type furrow opener (a) and press wheel for soil compaction (b)

Full size image

2.4 Working principle of RT

The developed system works on 12 V battery power supply. When the programme starts, the manipulator reset itself and comes to its initial position and stops. Once the start switch has been put on the PLC, the manipulator advances towards the seedling nearest to it of the first row to be extract first. After extracting the seedling, it moves back to its original position and the picked up seedling gets released into the delivery pipe. The IR sensor attached on the delivery pipe identifies the dropped seedling and then activate the DCU. The DCU actuates the dc motor attached on the wheels and the robot advance forward. The system is repeated till the last seedling on the tray was release. The complete system of the RT was shown in Fig. 6.

Fig. 6
figure 6

Block diagram showing working principle of the system: a Programme initiation, b Seedling extraction, c Robot movement

Full size image

2.5 Evaluation of test setup

A test set up of 7.5 × 0.62 m was developed to evaluate the developed transplanter under simulated condition (Fig. 7). The developed transplanter was tested with chilli as well as tomato seedlings of 30 days old and performance data was recorded. The success rate, leakage rate, as well as successful tranplanting are estimated as per the investigators [7,13, 23]:

$${\text{S }} = \, \left( {{\text{N}}/{\text{ N}}_{0} } \right) \, \times { 1}00 \, \%$$
(1)
$${\text{L }} = \, \left[ {{1} - \, \left( {{\text{N}}_{{1}} /{\text{N}}_{0} } \right)} \right] \, \times { 1}00 \, \%$$
(2)
$${\text{T}}_{{\text{S}}} = \, \left( {{\text{F}}/{\text{N}}_{0} } \right)_{{}} \times { 1}00 \, \%$$
(3)

where, S is overall seedling extraction success rate; L is the leakage rate; TS is the seedlings successfully transplanted; N0 is the total number of nursery seedlings in the nursery tray; N is the number of seedlings successfully extracted from the nursery tray; N1 is the total number of seedlings dropped into the furrow; F is the number of seedlings that were successfully transplanted into the furrow with seedling inclination of ≤ 300 and proper soil compaction.

Fig. 7
figure 7

Laboratory setup of the seedling transplanter (all dimension are in mm)

Full size image

The developed transplanter was evaluated in the laboratory thus factorial completely randomized design (FCRD) was selected at two levels of vegetable and two levels of transplanting methods as treatments. Test was carried out with two seedlings types viz. chilli and tomato with developed RT. Also, the manual method (control) of transplanting seedlings was considered to compare it with the developed system. All experiment was replicated thrice. All the recorded data were analyzed by the analysis of the Tukey’s test for comparisons of different treatment combinations with control was carried out using SAS 9.3 (SAS Institute Inc., Cary, NC, USA) at a 5% significance level. The coefficient of variation (CV) was determined to express the uniformity of distribution.

3 Result and discussion

3.1 Properties of nursery seedlings

The weight, height, stem diameter, and critical canopy diameter of the nursery seedlings were 13.1–13.4 g, 115.5–125.5 mm, 0.1–015 mm, 68–70 mm, and 400–450 mm, respectively. The optimum drop shatter height with minimum shattering of plug media for both seedlings was ranged from 400 to 450 mm. For tomato and chilli seedlings, the critical canopy diameters were 70.4 mm and 68.2 mm, respectively.

3.2 Software system

To extract the seedling and transplanting at a particular point, a transplanting device was developed using mechatronics. Here, a Microchip was used for the functioning of the driver of the stepper motor and DC motor (Fig. 8). For both DC and stepper motors, the microcontroller develops Pulse Width Modulation (PWM) signals with various pulse widths. The electronic switches are used to control the movement of the manipulator in XY plane. Once the start switch has been put on, the manipulator stops at the location and the gripper opens up to release the seedling. The stepper motor in the manipulator rotates clockwise and anticlockwise to pick the seedling when it grasps the seedling in its jaw. Both drivers get a 5 V DC power source from the microcontroller. The power source for both the motors and the microcontroller is a 12 V DC battery.

Fig. 8
figure 8

Components for the developed seedling extraction system using PLC

Full size image

3.3 Performance of the developed system

The developed transplanter was evaluated under simulated laboratory condition using chilli and tomato seedlings. (Fig. 9). Chilli seedlings of the Pusa Jwala variety and tomato seedlings of Abhilash variety were selected for the study. The development of the transplanter prototype took into account various physical and engineering characteristics of plug seedlings [40]. Both the vegetable seedlings were raised in seedling nursery trays of 96 cells with recommended media composition in a ratio of 2:1:1:1 (coco-peat, soil, vermiculite and perlite). Normally, nursery seedlings of 30 days having 4–6 leaves were selected for the study.

Fig. 9
figure 9

Success rate and successful transplanting of tomato and chilli seedlings

Full size image

The nursery tray containing seedlings was positioned on the seedling tray platform and the ASE was actuated by providing the set values showed in the liquid crystal display (LCD) screen. Once the start switch was engaged, the transplanting process initiated. The indices viz. success rate (Eq. 1), leakage rate (Eq. 2), and successful transplanting (Eq. 3) were determined and their results were presented in Table 2. The average success rate, leakage rate as well as successful transplanting for chilli seedling transplanting were 95.1%, 7.6% and 90.3%, respectively. In contrast, for tomato seedling transplanting, these values were measured at 91.0%, 13.9% and 86.8%, respectively. As evident from Table 2, mean success rate decreased by 4.1% for tomato seedling, which can be attributed to differences in their cultivation and physical characteristics when compared to chili seedlings. The success rate for both the crops ranged from 91 to 95% which is above 90% transplating efficiency that is considered as very good [8, 11]. While leakage rate and successful transplanting rate were reduced to 6.3% and 3.5% respectively due to tomato seedlings due to their lower sturdiness and straightness compared to chili seedlings. In the same trend, the seedling picking rate for 30-day-old chili seedlings stood at 95.1%, while the leakage rate was recorded at 7.6% [7].

Table 2 Performance evaluation of the developed transplanter with nursery seedlings
Full size table

The significance of crop and transplanting method on transplanting parameters is shown in Table 3. Transplanting methods were found to be significantly different (p < 0.05) for the transplanting parameters except plant mortality (Table 3). It means that plant mortality was same as compared to the manual method for the transplanting of any vegetable.

Table 3 Significance of crop and transplanting method on transplanting parameters
Full size table

The test data collected on seedling performance was then statistically analyzed for FCRD. The success rate was significantly (p < 0.05) affected by the change in crop seedlings and found 4.1% higher for chilli in comparison to tomato. Khadatkar et al. [12] obtained similar outcomes, reporting a transplanting efficiency of 92.6% for one crop and 90% for another. Notably, they observed that the transplanting efficiency was comparatively higher for the crop with sturdier seedlings, such as chili, as opposed to the more delicate tomato seedlings [12]. This is reflected due to the reason that the chilli seedlings are more sturdy and easy to handle, whereas as tomato seedlings are delicate and easily damaged. Similarly, success rate significantly (p < 0.05) contributed to higher (4.1%) successful transplanting for chilli followed by tomato (Fig. 9).

From Fig. 10, it can be observed that plant to plant spacing is significantly (p < 0.05, Table 3) higher in manual transplanting for tomato crop while for chilli crop statistically (p > 0.05) same. This may be attributed due to the fact of crop geometry and easy to notice the plant to place by manual method. The planting depth was significantly (p < 0.05, Table 3) higher (8–12 mm, Fig. 10) with transplanter as compared to manual (37–41 mm) transplanting for both the crops due to uneven surfaces whereas it was easy to dig hole manually. Miss planting was found to significantly (p < 0.05, Table 3) higher in transplanter in comparison to manual method while it was within the inacceptable limit (5–5.5%, Fig. 10). However, miss planting was non-significant for both the crops (Table 3). The average miss planting rate was reported by Khadatkar et al. [12] 4.5% for chili seedlings and 5% for tomato seedlings using automatic transpalnter. The higher miss planting rate observed for tomato seedlings can be attributed to their soft and delicate nature, making them more susceptible to damage during the planting process. Overall, successful transplanting rate was significantly (p < 0.05, Table 3) lower in the transplanter for both the crops i.e. chilli (92%) and tomato (86%). Further, the mortality rate was statistically (p > 0.05, Table 3) same in both the crops and both transplanting methods. This may be due to proper handling of seedling while extraction from the nursery tray and smooth release into the furrow with subsequent irrigation. Transplanting efficiency with transplanter was found to be non-significantly (p < 0.05) higher (5%) for tomato crop while significantly (p < 0.05, Table 3) higher (8%) in chilli crop when compared with manual transplanting.

Fig. 10
figure 10

Comparative values for various parameters under two crops against transplanter and manual method of transplanting (Tukey’s test for mean ± SD with same letter are not significantly different at p < 0.05)

Full size image

3.4 Advantages and limitations

This machine in the future may solve the problem of labour shortage during peak season, ensure timeliness of repetitive field operation, reduce drudgery, and also enhance work output are the major advantages. However, it also has some limitations viz. skilled operator is required, sensors and electronic part are susceptible to moisture and dust, repair and maintenance may not be easy, battery backup need to be higher duration for long hour operation.

4 Conclusions

Our study aimed to develop and test a mechatronic robotic transplanter for nursery-grown seedlings with an automatic seedling extractor. The central question we sought to address was whether this technology could effectively transplant seedlings with a 12 V battery supply, particularly focusing on chilli and tomato seedlings. The robotic transplanter demonstrated a high success rate of 91–95% for nursery-grown seedlings. This success rate underscores the potential of the technology to streamline the transplantation process. The automatic seedling extractor achieved a commendable seedling extraction rate of approximately 3 seedlings per minute with a single gripper. This rate can be further improved by incorporating multiple grippers. While the success rate was promising, we observed a leakage rate of 7.6–13.9%. This aspect suggests room for improvement in preventing damage or loss during the transplantation process. The leakage rate, though relatively low, is an area that warrants further refinement to minimize potential damage to seedlings during the transplantation process. Our work holds significant relevance and added value in the field of agriculture and horticulture. The successful development and testing of the mechatronic robotic transplanter and automatic seedling extractor can potentially revolutionize the way nursery-grown seedlings are transplanted. The use of a 12 V battery supply makes it a feasible and cost-effective solution for vegetable growers. This technology can find application in poly houses and shed nets, enhancing efficiency and reducing manual labor.

5 Directions for further research

Application of robotics in repetitive agricultural operations viz. vegetable transplanting increasing to achieve sustainability and nutritional food security. One such attempt was made here which has a limited transplanting rate. To achieve a higher transplanting rate, multi gripper system with multiple rows may be developed in the future. Again, it should be integrated with a vision-based system to identify healthy seedlings and empty cells as well as navigate in the right path with any human intervention. A multi-gripper type picking mechanism with multi-row planting may be developed to increase the field capacity of the developed system.