Two common tools were discussed: accelerometers and GPS-based sensors. The accelerometer monitors one of the most important aspects of the animals’ behavior, namely their acceleration, while GPS is capable of tracking their continued movements and locations, even when they are out of sight. This enables managers or farmers to know their exact position at any given time. There are many precision tools, like accelerometers, pedometers, GPS, and other motion sensors, which are used to monitor animals’ movements ^[1][33]. Recently, several devices equipped with accelerometers have been developed, such as SenseHub. The SenseHub monitoring technology is a complex yet user-friendly animal monitoring solution. It includes sensors to monitor vital signs or detect signs of illness. It allows for the unique identification of each tagged animal. This helps farmers to detect several parameters related to animal welfare, reproductive performance, health, and the nutritional status of individual animals and groups. The prospect of combining the accelerometer and GPS in studying animals’ behaviors will be discussed.

Sensors in general need to be attached correctly with the appropriate positioning, aiming to optimize the results by taking into account the animal’s well-being ^[2][34]. They are attached to some part of the body while aligned with its axes. This direct attachment prevents the sensor from moving independently of the animal, which may improve the recorded results ^[3][35].

As the name implies, the accelerometer is a device that measures acceleration, which is, by definition, the change in velocity over time ^[4][36]. Their utilization in monitoring animal behaviors started in the previous decade ^[4][36], and many studies have used this device to evaluate the physical activities of different species. The accelerometer can be distinguished based on the number of used axes, such as a 1-axis accelerometer that measures acceleration in a single direction, i.e., up or down; a 2-axis accelerometer that measures acceleration in two perpendicular directions, typically up/down and left/right; and a 3-axis accelerometer that assesses acceleration in all three dimensions, providing a more comprehensive view of the movement in 3-dimensional space ^[5][37]. This can help to sense orientation, coordinate acceleration, vibration, shock, and falling in a resisting medium ^[5][37]. In its application in monitoring animals, this tool could be attached to different parts of the body ^[6][38], and it can detect distinct changes and variations in the animal’s patterns and status ^[7][39].

Accelerometers are mostly used in recording animals’ daily activities, such as walking, ruminating, and lying, at different times of the day, seasons, and conditions for different species, with each study focusing on a set of behaviors, aiming to be as precise as possible. Studies tend to validate their results using visual observation by experts. Results usually vary from moderately to highly accurate. Cattle and the changes in their behavior have been the topic of several studies. Riaboff et al. ^[8][40] were successful in predicting cattle behaviors (grazing, walking, ruminating lying, ruminating while standing, resting while lying, and resting while standing) using a neck-collar accelerometer in a pasture-based system, with accuracy of 98%. Rumination behavior was monitored with an ear-tag accelerometer in a semi-enclosed barn. This behavior was detected with 98.4% accuracy ^[9][41]. Another type of behavior was monitored with a neck-collar and ear-tag accelerometer in an intensive system, which considered licking, where the overall performance of both types was acceptable (88 and 98% in accuracy) but with a small advantage in favor of the neck collar ^[10][42]. Calves’ behaviors were also monitored, such as suckling behaviors. Kour et al. ^[11][18] successfully identified and estimated more than 95% of suckling bouts and durations in a pasture-based system.

Another important matter is cattle health and welfare. Jaeger et al. ^[12][43] aimed at assessing cattle’s welfare under a normal production system (rotational grazing scheme) with an ear-tag accelerometer, which was found to be impacted by many factors, such as hygiene, aggressiveness, basic behaviors, and intra-herd rank. Lameness can affect cattle behaviors according to Thorup et al. ^[13][19], who used a leg-mounted accelerometer to prove that lame cows in intensive systems tend to spend more time lying down and less time walking. In a rotational grazing system, Tobin et al. ^[14][12] aimed at detecting illness before symptoms appeared, which was successful as they noticed a movement decrease in ill heifers 24 h previously. On the other hand, Sutherland et al. ^[15][44] considered diarrhea prediction with neonatal calves, noticing changes in behavior 4 days before the diagnosis.

The accurate detection of estrus and calving is very important for farmers; many researchers have attempted to predict them and observe changes in these periods. Benaissa et al. ^[16][45] used several combinations of accelerometers in a free-stall barn environment to detect estrus. This method was successful but was more accurate in the case of using one sensor on the same animal. A sudden behavior change could also indicate the time of calving. Borchers et al. ^[17][46] monitored cattle behaviors in a pasture-based environment and precisely detected these changes with sensitivity and specificity of more than 80%.

Comparing accelerometers’ accuracy, or even the validation of new ones by tested devices, has been the subject of numerous studies. In an intensive system, Borchers et al. ^[18][47] compared six commercially available accelerometers for different behaviors (lying, feeding, and rumination). Each sensor type was better adapted to some studied behaviors than others. An ear-tag accelerometer for the monitoring of calves’ drinking behavior was evaluated under an intensive system with accuracy of 96.2%. The early detection of changes in this behavior would prevent complications ^[19][48]. A noninvasive accelerometer for the monitoring of cattle sleep, attached to the harness, was evaluated, which was very accurate (92.2 ± 0.8%) ^[20][49]. The sensor’s position is also a variable that may affect its performance. Aloo et al. ^[21][50] studied this factor by placing the device between the dewlap, leg, and harness. They found that the latter was the most adequate. Van Erp-Van Der Kooij et al. ^[22][51] compared leg and neck placement and found a good correlation for both sensors for the studied behaviors (with a correlation coefficient of >0.85) except walking.

Sheep have been the subject of several extensive studies of behavior. Ikurior et al. ^[23][52] monitored sheep’s common behaviors using different accelerometer placements in an extensive system, and the overall accuracy was 89.6% for grazing, walking, and resting. Some specific behaviors may also be of interest, such as lying behaviors, which were successfully monitored by a leg-mounted accelerometer in an extensive system, and it has been concluded that many factors can affect sleep, namely sex, age, weight, and pregnancy ^[24][9]. Lamb suckling behaviors were investigated by Kuźnicka and Gburzyński ^[25][53] using a neck-mounted accelerometer; the detection rate was 95%. The monitoring of behaviors in relation to the diet, like chewing and biting activities, was achieved in a pasture-based environment, with sensitivity for biting and chewing activity (95.5% and 93%, respectively) improving as the time interval increased ^[26][54]. Monitoring the herd as a whole, by monitoring some animals and then using the data to predict or deduce the others’ behaviors and classify their behaviors with a neck-mounted accelerometer in a rectangular field, had a success rate of 74.8% ^[27][55], which is reasonable given that it is not possible or convenient to study every animal in the herd.

An important task is to evaluate the effectiveness of the sensor on sheep. A comparison between three types of accelerometers with variable configurations was conducted in a pasture-based environment. It was found that behaviors were successfully identified, with the best performance for the ear-mounted device (86% to 95% accuracy) ^[28][56]. Another sensor was evaluated to determine which behaviors could be detected easily. In a semi-improved pasture, the authors placed the sensor under the jaw and found that the detection of grazing behavior was the easiest. Another factor that alters the results is the placement of the sensor. Decandia et al. ^[29][57] considered three different placements (mouth, nape, and neck) under an extensive system. The neck-mounted device had the best results (90% of accuracy). Various sensors were also evaluated in recording some specific behaviors—for instance, urination events—with an accurate estimation rate of 100% ^[30][58], or lameness with three different placements (leg, ear, and neck), with the best results (87% of accuracy) for the leg deployment ^[31][13].

In the livestock industry, most research seems to be focused on reproduction and, therefore, parturition and lambing. Gurule et al. ^[32][59] studied the variations in ewes’ behaviors around parturition, in an intensive system, and achieved the monitoring of activity with 87.2% accuracy; the system was very helpful in predicting the approach of lambing. A similar study, but in an extensive system, showed that, around lambing time, grazing decreased in favor of other behaviors, such as lying and being active ^[33][60]. Concerning the sexual activities of rams, in a pasture-based environment, mounting and service detection were monitored successfully, with overall sensitivity of 77.9% ^[34][17].

Accelerometers, as sensors, offer valuable insights into various aspects of ruminant behavior, health, and management; nevertheless, it is important to mention certain challenges associated with accelerometer use, such as the data processing complexity, device attachment considerations, and battery life. Farmers, as end users, need to approve and accept the technology, which, in general, can be complex as they tend to prefer the visual approach ^[35][61]. According to a study by Van De Gucht et al. ^[35][61], farmers were reluctant to use an automatic lameness detection system, but this changed when they were informed about the serious consequences of lameness. Farmers’ interpretation and evaluation of data is a strategic procedure that aims to lead to informed decisions and improve overall livestock management. Key steps include ensuring the accuracy and quality of various sensor data, understanding the key capabilities of PLF technology, and setting benchmarks and goals for performance improvement. Farmers analyze patterns and trends over time, identify correlations between variables, and integrate PLF data into their farm management systems. Using PLF data to support decision making, farmers can adapt their practices, such as reproduction and welfare practices, while continuous training ensures effectiveness. A commitment to constant improvement with regular evaluation is necessary to be able to adapt to future demands.

A GPS sensor is a device that can be placed on different parts of the body to track and record animals’ real-time locations, and thus their movement, especially in large pastures ^[6][38]. This information on the animal’s position could provide details about the topography, vegetation type, water source locations, grazing locations, calving sites, and temperature ^[36][37][1,62]. GPS is especially useful in large pastures. The position and other data are communicated to a user’s server via global satellites. Therefore, one of its main uses is in monitoring livestock behaviors under pasture conditions. Several authors ^[38][39][40][63,64,65] have studied different aspects of cattle behavior. Castillo-Garcia et al. ^[41][25] evaluated the sheep’s grazing effects on vegetation to determine whether they were beneficial or not for pastures.

The cattle diet was monitored by Orr et al. ^[42][66], by tracking their preferred paths. They concluded that cattle favored shorter, easy-to-digest material during the day, while they selected material with higher crude fiber in the evening.

Social interaction with other species was examined by Brown et al. ^[43][67]. They studied the influence of the presence of cattle on the behavior of bighorn sheep. The latter stayed vigilant in the presence of cattle, with a noticeable decrease in foraging bite rates.

Ganskopp ^[44][3] manipulated the water and salt distribution, in a very large pasture, to determine which was more important to cattle. He found that water was more important, as they shifted towards it whenever they moved, which can be a very effective way to alter the cattle distribution with minimal interference.

There are also studies interested in developing GPS-based systems, such as Halasz et al. ^[45][68], who provided near-real-time monitoring with a constructed GPS tracking collar.

These studies and others clearly show the limited potential of GPS. Position data are indeed very helpful in many aspects but combining them with other sensors will allow their full potential to realized. This may include accelerometers, which will be the subject of the next section.

The combination of accelerometers and GPS increases the accuracy and sensitivity of detailed animal behavior detection ^[37][62]. Combining positioning data with other sensors, in general, would provide higher prediction accuracy as different behaviors emerge in different locations. Even with the same movement patterns, the animal’s real behavior could be deduced ^[46][69]. The purpose of the combination of different sensors is to obtain higher accuracy in detecting behaviors. For instance, Cabezas et al. ^[47][70] successfully classified cattle behaviors with a GPS and accelerometer integrated sensor attached to the neck in a pasture-based environment. The results were quite accurate, with grazing having the highest accuracy (93%) and ruminating (88.1%) having the lowest.

Goat behavioral classification, in an extensive system, was performed during different seasons with this combination, which confirmed that this ruminant tends to spend more time grazing during the spring but travels greater distances during the summer and autumn ^[48][23]. The set of data collected from two types of sensors may lead researchers to unexpected results, like Tobin et al. ^[49][71], who concluded, while surveying water malfunctioning based on cow behaviors, that the ones that experienced water shortages due to failure tended to stay closer to the water source.

This type of association between sensors is not fully understood. Researchers are still evaluating and testing their optimal use. Sprinkle et al. ^[50][72] tested this combination in a pasture-based environment, and patterns of grazing behaviors were accurately identified. In another study, the authors constructed a GPS collar combined with a three-axis accelerometer, and they tested this tool on steers. They had accurate results concerning grazing locations and timings, which tended to be during the morning and evening for 8.67 to 10.49 h per day ^[51][73].

There are also some studies of specific behaviors, such as Barker et al. ^[52][74], who developed a position- and activity-based system that detected lame cows successfully (accuracy ranged from 80.8% to 94.2%) based on changes in normal behavior. Another application was developed to detect sheep’s parturition. Fogarty et al. ^[53][15] used a global navigation satellite system (GNSS) tracking collar and an accelerometer ear-tag in their study. They had moderate accuracy at first, but it could be increased to 91% if an earlier false alert was permissible.