|
Safety Risk Assessment of an Autonomous Agricultural Machine
Guy Roger Aby1, Salah F. Issa1,*, Girish Chowdhary1
Published in Journal of Agricultural Safety and Health 30(1): 1-15 (doi: 10.13031/jash.15756). Copyright 2024 American Society of Agricultural and Biological Engineers.
1 Agricultural and Biological Engineering, University of Illinois, Urbana, Illinois, USA.
* Correspondence: salah01@illinois.edu, salahfuadissa@hotmail.com
The authors have paid for open access for this article. This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License https://creative commons.org/licenses/by-nc-nd/4.0/
Submitted for review on 28 July 2023 as manuscript number JASH 15756; approved for publication as a Research Article and as part of the Safety for Emerging Robotics & Autonomous Agriculture Collection by Community Editor Dr. John Shutske of the Ergonomics, Safety, & Health Community of ASABE on 8 December 2023.
Highlights
Abstract. The goal of this study was to analyze the safety implications of an autonomous agricultural machine (TerraPreta) using the standards ISO 18497 (ISO, 2018) and ISO 12100:2012 (ISO, 2012), as well as to investigate the ergonomics associated with the use of the autonomous agricultural machine. First, three engineers involved in the robot's manufacturing process were asked to evaluate the robot's functionalities compliance with the applicable safety standards and protective measures outlined in standard ISO 18497 (ISO, 2018). Second, while the robot was planting cover crop seeds, an attempt was made to identify and evaluate every risk connected to the robot using the risk assessment techniques outlined in ISO 12100:2012 (ISO, 2012). (1) Half (50%) of the functionalities of the autonomous agricultural machine complied with the safety requirements and protective measures described within the standard ISO 18497 (ISO, 2018). (2) The heavy reliance on past incident data of the risk assessment procedure described within the standard ISO 12100:2012 (ISO, 2012) makes it ineffective for new and revolutionary technologies such as autonomous agricultural machines where such data are not available. (3) Lifting a bag to fill the robot hopper with seeds was found to be a moderately hazardous activity associated with human-robot interaction. Multiple tentative solutions were provided to avoid this moderately hazardous activity.
Agriculture in the United States is not only one of the most hazardous industries, but also faces a constant and crucial concern for safety (Evans and Heiberger, 2016). Autonomous agricultural machines have the potential to increase safety in agriculture by removing the human operator from a mechanized system, reducing human exposure and associated risks that contribute to the high rates of injury associated with farm machinery operation.
Automation can reduce operational risk, but this does not mean that safety concerns are solved. Research conducted by Gerberich et al. (1998) on traditional farm machinery revealed that injuries frequently happen during repair, adjustment, and maintenance chores. It is conceivable that distinct hazards and situations will arise when a machine is used autonomously or with varying degrees of operator engagement. Risk analysis must take into account bystander exposures in a field, building, farmyard, or other places where an autonomous machine is functioning in addition to operator and repair/maintenance personnel dangers (Aby and Issa, 2023).
According to a recent study, more than 80% of autonomous agricultural machines are still in the development stage (Oliveira et al., 2021). Researchers now have a unique opportunity to address the safety of autonomous agricultural machines. Injuries and fatalities could result from failing to address the safety implications of new technology in a proactive manner. Consider a tractor as an example. The first tractors were developed in the 1890’s, and by the 1940’s, they were commonly found on US farms (Bellis, 2021). However, it was only in 1976 that injuries due to tractor rollovers or overturns were addressed through a mandate that agricultural tractors produced after October 25, 1976, shall be equipped with a rollover protection structure (ROPS)—(Murphy, 2023). Forty years later, nearly half of tractors on US farms do not have ROPS installed, and tractor rollovers remain a common source of injury on US farms (Loringer and Myers, 2008).
According to a recent study (Shutske et al., 2022), the International Organization for Standardization's (ISO) 12100:2012 (ISO, 2012) and 18497 (ISO, 2018) have been utilized to increase the safety of autonomous agricultural machinery. Therefore, the goal of this study was to determine whether ISO 12100:2012 (ISO, 2012) and 18497 (ISO, 2018) standards can be applied to effectively ensure the safety of autonomous agricultural machinery. Additionally, human-robot interaction was examined from an ergonomics standpoint.
Materials and Methods
Description of the Agricultural Autonomous Machine
The agricultural autonomous machine used in this study is a robot in development called the TerraPreta (shown in fig. 1). The robot is a cover crop seed hopper planter and can move between the field rows to sow seed. The TerraPreta is composed of a grain hopper atop a hub motor-propelled vehicle platform capable of four-wheel drive and powered by a lithium-ion battery. The grain is released through an auger and spreader and can cover a width equivalent to two standard 30-inch rows of maize. Sensors on the robot include embedded front and rear LiDAR, embedded IMU, two RGB cameras with gimbal stabilization, motor encoders, and embedded RTK GPS. The robot’s computer is an Intel NUC with an I7 processor, 16 GB of RAM, a 1 TB hard drive, and a Raspberry Pi 3. The robot receives instructions (jobs) through a regular standard tablet connected via WiFi (table 1). This study was conducted on an earlier version of the robot. Newer versions of the robot may have changes, including improvements, that would not be reflected in this article.
 |
| Figure 1. TerraPreta robot located at the entrance of maize field row. |
| Table 1. Summary of key specifications of the agricultural autonomous machine. | ||
| Semi-autonomous (level 2 Autonomy) | The robot moves autonomously within field rows, but a human must stay nearby in order to help the robot when it is not able to return to its intended path. Sometimes, the robot cannot automatically turn from one row to another, and the user needs to drive it manually, using the tablet, to position the robot at the entrance of the new row. | |
| Connectivity | WiFi | |
| Speed | The robot speed is 0.89 m/s (2.00 mph). | |
| Weight | 45 pounds (20.4 kg) when the hopper is empty. Around 100 pounds (45.36 kg) when the hopper is fully loaded with seeds. | |
| Module of decision-making autonomy of the robot. | 4 navigation modes: LiDAR row follower, Vision based row follower, GPS waypoint follower, and Full path Navigation | |
| LiDAR row follower: The robot uses the information from the LiDAR to scan the environment, and then the system uses those readings to determine the distance between the LiDAR and each row, and the robot heading. | ||
| Vision-based row follower: The robot uses the front camera to scan the environment, then uses a Resnet18-based Convolutional Neural Network (CNN) to determine the distances between the camera and each row and the robot’s heading. | ||
| GPS waypoint follower: For this system, the user needs to collect a set of waypoints along the desired path that the robot has to follow, and then the user needs to come back to the initial waypoint to start the autonomous navigation based on GPS information. | ||
| Full path Navigation: This is the combination of the vision/LiDAR and GPS navigation systems. When the robot is in a row, the system uses LiDAR/vision navigation, and when it reaches the end row, the system switches from LiDAR navigation to GPS to continue the path. | ||
ISO 18497 Agricultural Machinery and Tractors – Safety of Autonomous Agricultural Machines – Principles for Design
The International Organization for Standardization (ISO 18497) (ISO, 2018) provides principles for the safe design of autonomous machines and vehicles (agricultural tractors, tractor implement systems, implements, and self-propelled machinery). A comprehensive review of safety requirements and protective measures provided within the standard was completed. A total of 95 safety and protective measures were evaluated for their applicability to the TerraPretarobot. This list was further evaluated, and only guidelines that might address safety concerns for the robot were selected, summarized, and sent to the robot’s manufacturers as an internet-based survey questionnaire. The survey was designed as a simple yes/no questionnaire that the manufacturer used to assess the implementation of each safety guideline.
ISO 12100:2012_ Safety of Machinery – General Principles for Design – Risk Assessment and Risk Reduction
The International Organization for Standardization (ISO 12100:2012) (ISO, 2012) has been extensively used to improve the safety of non-autonomous agricultural machinery by presenting procedures for assessing risk (Sandner, 2021). ISO 12100:2012 (ISO, 2012) considers the risk of a specific hazard to be a function of the severity of harm resulting from the hazard and the probability of the occurrence of that harm. The TerraPreta robot was used to plant seeds at the Illinois Autonomous Farm (IAF), located in Urbana, Illinois. During planting operations, following the risk assessment methodology described in ISO 12100:2012, an attempt was made to identify and assess all hazards associated with the autonomous agricultural machine.
 |
| Figure 2. Angle of load (A), Horizontal (H), Coupling (C), Distance travelled (D), Vertical location (V), and Frequency (F) are the variables used in measuring the recommended weight limit (NIOSH, 2021). |
Ergonomics
While the TerraPreta robot was used to plant seeds at the IAF, all human-robot interactions during operations, maintenance, and cleaning were identified and documented. Each human-robot interaction was reviewed for potential ergonomic implications. The frequency and type of movement were documented for each interaction. If there was any lifting required, the Revised NIOSH Lifting Equation (RNLE) (CCOHS, 2023) was utilized to determine the extent of any potential risk to the user. The RNLE is considered one of the most widespread tools used to assess and design manual lifting tasks among all ergonomic risk assessment tools. The RNLE is used to calculate the Recommended Weight Limit (RWL), which determines whether an object is too heavy for the task or not (variables for RWL are shown and defined in fig. 2). Another output of the RNLE is the Lifting Index (LI) (Waters et al., 2021). A lifting index value of 1.0 or less is recommended for safe manual lifting tasks.
(1)
where
RWL = recommended weight limit (lbs)
LC = load constant (lbs)
H = horizontal location of the object relative to the body (in)
V = vertical location of the object relative to the floor (in)
D = distance the object is moved vertically (in)
A = asymmetry angle or twisting requirement (degrees)
F = frequency and duration of lifting activity (Lifts/min)
C = coupling or quality of the worker’s grip on the object.
(2)
where
W = weight of item to be lifted (lbs)
RWL = recommended weight limit (lbs).
Results and Discussion
ISO 18497 Agricultural Machinery and Tractors – Safety of Autonomous Agricultural Machines – Principles for Design
The International Organization for Standardization 18497 (ISO, 2018) contains 95 safety requirements and protective measures. The TerraPreta robot, based on its specific design and application, is subject to only 8 of the 95 safety requirements and protective measures. Failure to meet these applicable measures could raise safety concerns for the TerraPreta robot.
The TerraPreta robot is exempt from the irrelevant 87 safety regulations and precautions. The majority of ISO 18497's safety regulations and precautions apply to highly automated agricultural machines (HAAMs), which are mobile vehicles or machines that carry out tasks under the direct management of a control system without the need for direct input from a local or remote operator (ISO, 2018). The TerraPreta robot is currently a semi-autonomous machine, as it requires commands from a nearby or distant operator in order to function. The eight safety criteria and preventative actions chosen for this study apply to all agricultural robots, not only those meeting the description of a HAAM. Examples of the 87 safety requirements and protective measures that are not applicable to the TerraPreta robot are: (1) “Should a fault occur during highly automated operation, the HAAM shall stop motion automatically, and highly automated operation shall be disabled until the fault is rectified.” (2) “A pendant control shall be provided with the HAAM that does not have an operator on-board station.” (3) “When a highly automated operation is suspended because of object detection or a malfunction, appropriate additional warnings shall be given to indicate the cause.
Table 2’s findings show that the robot complies with 4 of the 8 safety requirements and protective measures listed in ISO 18497 (ISO, 2018). To determine whether the robot complies with the safety requirements, a total of three engineers who have worked on developing the robot had a group discussion before reaching consensus on one answer (yes or no). Four criteria were met, but four were not. Three of the four unmet criteria included ISO-guided requirements related to identification, notification, and response to obstacle detection. This is an important safety function, as multiple hazards exist in agricultural settings, including people, animals, vehicles, holes, ditches, power lines, trees, and uneven ground (Reina et al., 2016; Aby and Issa, 2023). A safe autonomous agricultural machine must be able to robustly detect and avoid all of these obstacles, especially when operated without human supervision. At this stage in its development, the TerraPreta robot cannot be operated without an individual in the field monitoring the robot. It is important to note that the robot design and features continue to evolve in newer versions of the product, and future plans include adding algorithms to enable it to accurately recognize and avoid obstacles in an agricultural context. An additional concern for autonomous agricultural machines in general is the effectiveness of ISO 18497 (ISO, 2018) with regards to obstacle detection procedure. Currently, the obstacle defined in ISO 18497 (ISO, 2018) is a barrel-shaped device with the intended purpose of simulating a human seated position with dimensions of 80 cm in height and 38 cm in width (fig. 3). Steen et al. (2016) developed a deep convolutional neural network (DCNN) to recognize the barrel according to the standard’s requirements. The DCNN capacity to recognize the ISO obstacle in grass cutting and row crops was evaluated. The ISO obstacle was detected with a precision of 99.9% in row crops and 90.8% in grass mowing, while simultaneously not detecting individuals or other highly distinctive obstacles, such as animals in the image (Steen et al., 2016). The software used by the robot was not examined in this study.
| Table 2. Results of compliances between the functionalities of the agricultural autonomous machine and the safety requirements and protective measures described within the standard ISO 18497 (ISO, 2018). | |||
ISO 18497 (ISO, 2018) | Answers | ||
| Yes | No | ||
| Is the robot equipped with a perception system capable of detecting and locating persons or other obstacles relative to the robot? |  | ||
| Is the robot equipped with a perception system capable of locating and positioning itself? |  | ||
| While performing autonomous operations, is the robot capable of giving an audible or visual alarm when an obstacle is detected? |  | ||
| Can the robot be started or stopped remotely by an operator? |  | ||
| Can the robot be adequately supervised remotely during working conditions? |  | ||
| Is the robot sufficiently protected against unintentional actuation? |  | ||
| Is the robot sufficiently protected against failure to stop remotely? |  | ||
| Is the robot equipped with a perception system capable of confirming that the hazards zone is obstacle-free before any operations can be initiated? |  | ||
ISO 12100:2012 Safety of Machinery – General Principles for Design – Risk Assessment and Risk Reduction
Six hazards were identified while observing the robot while it was in operation. These hazards included unexpected moves, failure to stop, unexpected changes of direction, the wrong direction at start-up, the unintentional start of the spreader, and the unintentional start of the robot auger. The respective hazardous situation, hazardous event, and possible harm were noted in table 3.
| Table 3. List of all hazards identified, including their respective hazardous situation, hazardous event, and possible harm. | ||||||
| Hazard | Hazardous Situation | Hazardous Event | Possible Harm | |||
| Unexpected move | User located nearby the robot | Struck by powered vehicle—robot | Surface wounds and bruises | |||
| Failure to stop | User located in the robot direction | Struck by powered vehicle—robot | Surface wounds and bruises | |||
| Unexpected change of direction | User located in the robot direction | Struck by powered vehicle—robot | Surface wounds and bruises | |||
| Wrong direction at start-up | User located in the robot direction | Struck by powered vehicle—robot | Surface wounds and bruises | |||
| Unintentional start of robot spreader | Robot hopper containing seed and user face located nearby the robot | Struck by thrown objects—seed | Pain, eye injury | |||
| Unintentional start of robot auger | Finger of the user being close to the auger | Caught in running equipment or machinery | Entanglement | |||
For the first four hazards (unexpected movement, failure to stop, unexpected change of direction, and wrong direction at start-up), the only potential hazardous event is being struck by the robot. The robot, built from rubber and steel materials, weighs approximately 20.41 kg (45 lbs) when the hopper is empty and 45.36 kg (100 lbs) when the hopper is full of seeds. The robot’s top speed is 3.22 kph (2.00 mph, or 0.89 m/s). Determining whether the impact could bruise or even fracture bones can be determined by calculating the force and energy from the impact collision as follows:
 |
| Figure 3. Schematic of the barrel-shaped object that is defined as an obstacle in ISO 18497 (ISO, 2018) with measurements in cm. |
(3)
(4)
(5)
where
F = force (N)
E = energy of impact (kJ)
Edensity= energy density (kJ/m2)
m = mass of the object (kg)
a = acceleration (m/s2)
?v = change in velocity at impact (m/s)
S.A. = surface area of impact (m2).
The duration of the impact varies depending on whether the person recoils and where and how the impact occurred. One study found the impact speed of slapping a rubber block from a distance of 10 cm ranged from 0.0098 to 0.0193 seconds (Adamec, 2020). Using these values for impact duration, the machine could produce an impact force ranging from 2.1 kN to 4.1 kN. Nahum et al. (1968) found that the minimum tolerance for a skull bone to fracture ranged from 2.2 kN for the cheekbone to 11 kN for the frontal bone (Nahum et al., 1968) with a contact area of 6.5 cm2. In another study, Sugiura et al. (2019) reported that porcine extremities might bruise at energy densities greater than 9.9 kJ/m2 and human calf might bruise at energy density of 17 kJ/m2 (Sugiura et al., 2019; Desmoulin and Anderson, 2011). Based on the robot's weight and max speed, it could produce a maximum of 18.12 J in energy. This means the area of impact needs to be smaller than 10.7 cm2 to bruise and injure a person (using 17 kJ/m2 as a base estimate). Given these values and the most likely impact location is the rubber hopper, the most probable outcome is discomfort, pain, or bruising, though more severe injury can occur based on impact surface area, impact duration, and the body part impacted. The robot could also be more harmful to a child or pet.
For the fifth hazard (unintentional start of robot spreader), the potential hazardous event is the seeds ejected on a trajectory toward an individual’s face. The robot can spread seed quickly and over a large area (1.2-1.5 m diameter). The seeds are shot out at an upward trajectory. During field observations, workers were frequently found to be close to the robot when seed spreading operations began. These seeds could potentially hit an individual’s face and possibly eyes and could result in a serious injury to the eyes if an individual is not wearing appropriate ,well-fitting PPE. However, as the machine’s autonomy continues to evolve, fewer, if any, workers might be expected to be near the robot during operations.
For the sixth hazard (unintentional start of robot auger), the potential hazardous events are amputation of fingers and entanglement of hand(s). The auger is located at the bottom of the seed hopper and is not accessible from the outside. During repairs or seed filling, an operator could potentially reach the auger with their finger(s), and an entanglement could occur. If the operator is wearing gloves, the risk of entanglement might be even greater.
The risk can be divided into three categories, according to the risk assessment techniques outlined in the ISO 12100:2012 standard (ISO, 2012): (1) Slight, (2) Serious, and (3) Death. The first four hazards (unexpected move, failure to stop, sudden change of direction, and improper direction at start-up) were each rated as having a slight degree of severity because the most likely outcome would be pain or bruising. As was already indicated, there is a good possibility that any incidents would only result in discomfort or bruising due to the impact area (rubber hopper), robot low speed, and impact surface area. However, in certain circumstances, incidents can lead to ankle/knee sprains and even bone fractures. The last two risks, the unintended activation of the robot auger and spreader, have a serious level of severity because a severe eye injury or entanglement might occur.
The probability of occurrence is an essential element in the risk assessment process. The ISO 12100:2012 standard (ISO, 2012) states that the likelihood of an event depends on the machine's probable applications or misuses, accident history, typical design considerations, reliability, and other statistical data. However, because the TerraPreta robot is a brand-new technology, there is no prior information on it. Therefore, using the risk assessment technique outlined in ISO 12100:2012 (ISO, 2012) to evaluate the risks connected with the TerraPreta robot was notably hindered by the reliance on pre-existing knowledge. With about more than 80% of autonomous agricultural machine being at the development phase (Olivera et al., 2021), this is a concern that needs to be addressed by industry. A recent survey found that only half of the 153 respondents working with autonomous agricultural machinery use ISO 12100:2012 (ISO, 2012) during the design, development, and testing of autonomous agricultural machinery (Shutske et al., 2022). This fact might provide support to the limitations of ISO 12100:2012 (ISO, 2012) in evaluating the risk of autonomous agricultural robotics.
Ergonomics
Cleaning of the Agricultural Autonomous Machine
The robot is only cleaned when its wheels are full of mud. The user waits for the mud to dry on the wheel, then uses pliers to cut off the mud.
Charging the Agricultural Autonomous Machine
The robot utilizes a swappable lithium-ion battery, accessed through a back panel in its body. To charge the battery, users simply remove it by unscrewing one screw from the robot and charge it separately.
Turning the Agricultural Autonomous Machine from One Row to Another
The robot is equipped with a control tablet that is used to send instructions to the robot. Commands are transmitted remotely from a control tablet to the robot to make it turn from one row to the next. The robot operator currently needs to be near the robot during operations due to low WiFi signal strength, and the robot lacks robust detection of obstacle systems. This includes staying in the fields for long hours under the sun. This is expected to change as the robot continues through its development cycle.
Filling the Agricultural Autonomous Machine Hopper with Seeds
The operator of the robot was observed to lift and hold a 22.68-kg (50 lbs) bag of seeds to fill the robot hopper with seeds (fig. 4). Lifting a 22.68-kg (50 lbs) bag of seed can cause long-term health concerns and several severe injuries, such as back sprains, muscle pulls, wrist injuries, and elbow injuries, under unfavorable conditions related to job design, worker posture, and other factors. Therefore, to evaluate the risk implications associated with this activity, the step-by-step procedure of the RNLE (Waters et al., 2021) was used in this study (fig. A1).
During the seed planting operations, measurements were obtained to calculate the WL and LI. As outlined in table 4, the findings indicate that a 22.68-kg (50 lbs) bag of seed was lifted at least once per acre. Moreover, the robotic system seed planting rate is one acre every 40 minutes. These results suggest that, under optimal conditions and utilizing a single robot, a 22.68-kg (50 lbs) bag of seed is lifted approximately once every 40 minutes.
 |
| Figure 4. A 22.68-kg (50 lbs) bag of seed is being lifted and held to fill the robot hopper with seeds. |
| Table 4. The total seed planted, planted hours, number of acres planted, and frequency at which the seed bag is lifted. | |||
| Total Seed Planted (kg) | Planted Hours (min/acre) | Acres Planted | Number of Times a 22.68-kg Bag Was Lifted to Fill the Robot Hopper (per acre) |
| 22.68 | 40 | 1 acre | 1 lift |
| 113.4 | 200 | 5 acres | 5 lifts |
| 136 | 360 | 6 acres | 6 lifts |
Step 1: Measure and record task variables
To obtain the following values, measurements were taken during seed planting operations according to the procedures described in a study conducted by Waters et al. (2021).
Load Constant (LC) = 51 lbs
Object Weight = 50 lbs
Hand Location at origin (H) = 12 in
Hand Location at destination (H) = 10 in
Vertical height at origin (V) = 0 in
Vertical height at destination (V) = 32 in
Vertical distance (D) = 32 in
Asymmetry Angle degrees at origin (A) = 0, (no twisting)
Asymmetry Angle degrees at destination (A) = 0, (no twisting)
Frequency Rate (F) = 0.025 Lifts/min. Less than 0.2 lifts/min, use 0.2 lifts/min according to the RNLE guidelines.
Duration = A total of 6 hours, but 1 hour will be considered for the calculation because the object was lifted once per hour.
Object Coupling C = Poor. This corresponds to 0.90 as numerical value.
Step 2: Determine the multipliers and compute the RWLs.
To determine the multipliers, different tables (Waters et al., 2021) were used (tables A1-A6).
LC = 51 lbs
HM (origin) = 0.83
HM (destination) = 1.00
VM (origin) = 0.78
VM (destination) = 0.98
DM = 0.88
AM = 1.00
FM = 1.00
CM = 0.90
RWL = (LC) * (HM) * (VM) * (DM) * (AM) * (FM) * (CM)
RWL (Origin) = 51 * 0.83 * 0.78 * 0.88 * 1.00 * 1.00 * 0.90 = 26.15 lbs
RWL (Destination) = 51 * 1.00 * 0.78 * 0.98 * 1.00 * 1.00 * 0.90 = 35.01 lbs
Step 3: Compute the Lifting Index (LI)
Origin: LI = Object Weight / RWL = 50 / 26.15 = 1.9
Destination: LI = Object Weight / RWL = 50 / 35.01 = 1.4
The weight being lifted (50 lbs) currently by the robot’s users is greater than the RWL at both the origin and the destination of the lifts (26.15 lbs and 35.01 lbs, respectively). The LI at the origin is 1.9, and the LI at the destination is 1.4. According to Fox et al. (2019), these values indicate that this lift would be moderately hazardous for the majority of healthy industrial workers. This task should be redesigned to reduce the lifting index.
The first option is to reduce the vertical distance between the origin and destination by either increasing the origin or lowering the destination of the lift. Lowering the lift's destination would be difficult because the robot's height would have to be reduced, which would require changing the robot's design. However, raising the origin of the lift can be done by always making sure the seed bag does not lay on the floor but rather on a chair or table. Additionally, the coupling between the worker's hand and the bag needs to be of higher quality. As opposed to a poor coupling, which often requires greater maximum grasp forces and lower permissible weight for lifting, a good coupling will minimize the maximum grasp forces needed and raise the acceptable weight for lifting. With items that the hand can readily wrap around, such as a well-designed container, a nice connection can be formed. However, poor coupling happens when the lifted object is difficult to handle, such as bags. Therefore, using seed containers rather than seed bags is preferred. Moreover, reducing the weight in half (25 lbs), would reduce the LI substantially without affecting the lifting equation multiplier since the frequency rate would still be less than 0.2 lifts per minute.
Conclusions
The study aimed to analyze the safety of an autonomous agricultural machine using two distinct standards of the International Organization for Standardization (ISO 18497 (ISO, 2018) and ISO 12100:2012 (ISO, 2012)), and human-robot interaction was examined from an ergonomics perspective. The main findings and inferences include:
- Half of the safety and protective measure functionalities of the autonomous agricultural machine investigated complied with the safety requirements and protective measures described within the standard ISO 18497 (ISO, 2018).
- The risk assessment processes outlined in ISO 12100:2012 (ISO, 2012) cannot be utilized to effectively assess and reduce the risk associated with autonomous agricultural equipment since the risk assessment procedure requires prior knowledge about the autonomous machine, which is generally not available. This standard might also be of limited use for new technologies.
- The activity of lifting seed bags to put seeds in the hopper of the autonomous agricultural machine was found to be moderately hazardous for the majority of healthy industrial workers. Ergonomics methods can be used to assess and improve the safety of autonomous agricultural machines.
Overall, the findings of this research reveal the need for continued development and evolution of standards and methods to further advance the safety of autonomous agricultural machines.
Acknowledgments
This research was supported with funding provided by Discovery Partners Institute (DPI) and the UIUC Agricultural Safety & Health Program. The authors are grateful to the staff of the Illinois Autonomous Farm (IAF) for their assistance with this project and for providing us with an autonomous agricultural vehicle. Open Access was supported by the SAFER AG project supported by the intramural research program of the U.S. Department of Agriculture, National Institute of Food and Agriculture, proposal # 2022-07106 / accession # 1029426. The findings and conclusions in this preliminary publication have not been formally disseminated by the U.S. Department of Agriculture and should not be construed to represent any agency determination or policy.
Appendix
Figure A1. Step-by-step procedure of the RNLE (Waters et al., 2021)
References
Aby, G. R., & Issa, S. F. (2023). Safety of automated agricultural machineries: A systematic literature review. Safety, 9(1), 13. https://doi.org/10.3390/safety9010013
Adamec, J., Schöpfer, J., Hofer, P. et al. Die Ohrfeige. Teil 1 – Allgemeines und Biomechanik. Rechtsmedizin 30, 1–7 (2020). https://doi.org/10.1007/s00194-019-00362-3
Bellis, M. (2021). American farm machinery and technology changes from 1776-1990. ThoughtCo. Retrieved from https://www.thoughtco.com/american-farm-tech-development-4083328
CCOHS. (2023). NIOSH Lifting Equation (revised) - NIOSH Lifting Equation - Assessing Relevant Handling Factor. Canadian Center for Occupational Health and Safety. Retrieved from https://www.ccohs.ca/oshanswers/ergonomics/niosh/assessing.html
Desmoulin, G. T., & Anderson, G. S. (2011). Method to investigate contusion mechanics in living humans. J. Forensic Biomech., 2.
Evans, J., & Heiberger, S. (2016). Agricultural media coverage of farm safety: Review of the literature. J. Agromed., 21(1), 91-105. https://doi.org/10.1080/1059924X.2015.1106376
Fox, R. R., Lu, M.-L., Occhipinti, E., & Jaeger, M. (2019). Understanding outcome metrics of the revised NIOSH lifting equation. Appl. Ergon., 81, 102897. https://doi.org/10.1016/j.apergo.2019.102897
Gerberich, S. G., Gibson, R. W., French, L. R., Lee, T.-Y., Carr, W. P., Kochevar, L.,... Shutske, J. (1998). Machinery-related injuries: Regional rural injury study — I (RRIS — I). Accid. Anal. Prev., 30(6), 793-804. https://doi.org/10.1016/S0001-4575(98)00032-3
ISO. (2012). ISO 12100:2012: Safety of machinery - General principles for design - Risk assessment and risk reduction.
ISO. (2018). ISO 18497.2: Agricultural machinery and tractors - Safety of autonomous agricultural machines - Principles for design.
Loringer, K. A., & Myers, J. R. (2008). Tracking the prevalence of rollover protective structures on U.S. farm tractors: 1993, 2001, and 2004. J. Saf. Res., 39(5), 509-517. https://doi.org/10.1016/j.jsr.2008.08.003
Murphy, D. (2013). OSHA ROPS and operator instruction requirements. Penn State Extension. Retrieved from https://extension.psu.edu/osha-rops-and-operator-instruction-requirements
Nahum, A. M., Gatts, J. D., Gadd, C. W., & Danforth, J. (1968). Impact tolerance of the skull and face. Proc. 12th Stapp Car Crash Conf. SAE International. https://doi.org/10.4271/680785
NIOSH. (2021). Canadian Center for Occupational Health and Safety.
Oliveira, L. F., Moreira, A. P., & Silva, M. F. (2021). Advances in agriculture robotics: A state-of-the-art review and challenges ahead. Robotics, 10(2), 52. https://doi.org/10.3390/robotics10020052
Reina, G., Milella, A., Rouveure, R., Nielsen, M., Worst, R., & Blas, M. R. (2016). Ambient awareness for agricultural robotic vehicles. Biosyst. Eng., 146, 114-132. https://doi.org/10.1016/j.biosystemseng.2015.12.010
Sandner, K. J. (2021). The efficacy of risk assessment tools and standards pertaining to autonomous agricultural machinery. MS thesis. University of Wisconsin, Department of Biological Systems Engineering.
Shutske, J. M., Sandner, K. J., & Jamieson, Z. (2022). Risk assessment methods for automated agricultural machines: Current practice and future needs. Paper No. 2201102. Proc. 2022 ASABE Annu. Int. Meeting. St. Joseph, MI: ASABE. https://doi.org/10.13031/aim.202201102
Steen, K. A., Christiansen, P., Karstoft, H., & Jørgensen, R. N. (2016). Using deep learning to challenge safety standard for highly autonomous machines in agriculture. J. Imag., 2(1), 6. https://doi.org/10.3390/jimaging2010006
Sugiura, R., Fujikawa, T., Nishikata, R., & Nishimoto, T. (2019). Soft tissue bruise injury by blunt impact in human-robot interaction - difference of tolerance between chest and extremities. Proc. 2019 19th Int. Conf. on Control, Automation and Systems (ICCAS) (pp. 792-797). IEEE. https://doi.org/10.23919/ICCAS47443.2019.8971656
Waters, T. R., Putz-Anderson, V., & Garg, A. (2021). Applications manual for the revised NIOSH lifting equation. Centers for Disease Control and Prevention, National Institute for Occupational, Safety Health. https://doi.org/10.26616/NIOSHPUB94110revised092021