Psychophysical State Aspect during UAV Operations

Abstract

The development of unmanned aerial vehicles (UAVs) and the increasing air traffic of these devices make it necessary to pay attention to the issue of the human factor in UAV operations. In this article, tests were conducted in real conditions on the unmanned aerial vehicle operator’s (UAVO) psychophysical state during training. The parameters of the human cardiovascular system, and more specifically the heart rate variability (HRV), were used to conduct research and analysis. The purpose of this research is to elaborate the typical HRV parameters for student operators during UAVO training. These reference values could be used during UAVO training to assess candidates’ psychophysical state objectively and could allow for the monitoring of operators’ state and management of their cognitive load. Monitoring operators’ state may have a positive impact on increasing training effectiveness. Research confirmed the thesis that HRV parameters are significantly different during performed tasks with cognitive load and can be used to assess candidates’ psychophysical state objectively. This can help flight instructors perform student assessment, meaning that they would not have to rely only on their subjective feelings.

1. Introduction

Unmanned aviation is one of the fastest growing modes of transport, and having a pilot’s license is becoming increasingly popular. Since 2016, the number of unmanned aerial vehicle operator (UAVO) certificates issued in Poland has increased by 50% annually [1,2,3,4]. Flight safety is the most important aspect in air transportation [5] and unmanned aerial vehicle (UAV) traffic can be monitored and managed through three main elements: operators, UAVs and airspace [6]. One study presents a classification of UAV safety solutions that can be found in the scientific literature, but none of them are concerned directly with the human factor [7]. Another highlights the significant matter of the increasing need for skilled UAV operators and the problem of research on effective UAV operator training not keeping pace with demand [8]. An important finding was that student operators who were overconfident were 15 times more likely to crash [8]. With the knowledge that commercial UAV licensing is expected to significantly increase in the next few years, these results suggest more work is needed in the UAVO training field, and the methods used so far have to be developed. According to the literature, human error is the most difficult component to assess in the prevention of aviation accidents [9,10,11]. Assessments of the mental workload for UAV operations are essential in realising the causes of UAV accidents [12]. The development of UAVs, the increasing air traffic of such devices and the increase in the number of issued pilots’ certificates make it necessary to pay attention to the matter of the human factor in UAV operations. Providing appropriate safety and efficiency during different operations is possible when the operator performs each task at the right cognitive load level [13,14,15,16]. By analysing the legal regulations (at the level of the European Union and Poland) regarding training for UAVOs, the following problems were identified: the relationship between the assessment of a student operator’s predisposition and capacity to perform individual exercises based on the subjective assessment of an instructor; the subjective assessment of a student operator’s performance of individual exercises; the lack of objective methods, based on behavioural science, to assess the activities performed by student operators during training; and the small amount of time allocated to individual exercises. The legal regulations also do not provide explicit information about the possibility of adapting the training program to the individual needs and predispositions of the candidate.

It was decided to characterize the defined problems as those related to the effectiveness of UAVO training, which in terms of research concerns ensuring safety. How can we manage the cognitive load level and monitor the operator’s psychophysical state during training and operations? Human error is usually studied by behavioural and subjective measures. However, subjective and behavioural measures are known as overt measurements and are relatively limited, as they do not allow for continuous and direct assessment of cognitive processes [17]. On the other hand, objective physiological measures can be a good solution for performing psychophysical state assessment. In this research, objective methods (for example, electroencephalography, eye tracking, electrocardiography, etc.) are focused on. These types of measurements allow for analysis to be performed using specific quantitative values originating from human physiological parameters. Most research has applied the psychophysical state analysis method based on cardiovascular system work [18,19].

In this article, the decision was made to focus on the human factor in training for UAVOs in visual line of sight (VLOS) and to conduct tests in real conditions on the student operators. The parameters of the human cardiovascular system, and more specifically the heart rate variability (HRV), were used to conduct research and analyse psychophysical states. Based on the literature analysis, it can be stated that the work of the cardiovascular system reflects the ability of the human to handle situations with significant cognitive load [20,21,22,23,24]. Different research shows that the psychophysical state of a subject can be determined based on HRV parameters—which is possible because of the direct relationship between the parasympathetic and sympathetic nervous systems [25,26,27,28,29,30,31]. The main purpose of this research is to elaborate the typical time-domain and nonlinear HRV parameters for student operators during UAVO training. The calculation of typical HRV parameters during UAVO training will be possible and scientifically valid if differences in HRV parameters between rest measurement and cognitive load measurement are statistically significant. Taking this into account, we formulated the thesis that HRV parameters are significantly different during the performance of tasks with cognitive load than they are at rest. These reference values could be used during UAVO training to assess the psychophysical state of candidates objectively and could allow for the monitoring of operators’ state and management of their cognitive load. Knowing about changes in student operator psychophysical state during training could help instructor with problematic exercise identification. This way of monitoring candidates for operators may have a positive impact on training effectiveness and safety. This approach could limit the number of aviation accidents and could be the next development in UAV safety.

2. Materials and Methods

Tests were carried out in real circumstances, during the training for a UAVO license under visual line of sight (VLOS) for flights with the National Standard Scenario NSTS-01. This type of license allows operators perform operations in VLOS of a UAV weighing no more than 4 kg. Conducting research in real conditions is a rare approach to research on the analysis of psychophysical states. Typically, researchers conduct research in simulated conditions due to the possibility of controlling many parameters related to the performance of operations and limiting the influence of various external factors, e.g., weather. Due to the specific nature of the research, i.e., analysis of candidates’ psychophysical condition and determination of parameters characterizing an operator’s psychophysical condition, it was decided to conduct research in real conditions. Thanks to this, the obtained parameters are adequately comparable to those occurring during typical training of future operators. The research incorporated consultation with an instructor who is responsible for training future operators on a daily basis, and his comments were taken into account during the research. All measurements were performed under the supervision of an instructor. Additionally, an opinion was presented confirming the validity of the research and the fact that knowing a candidate’s psychophysical condition during training exercises would make it much easier to make decisions related to further training and allowing the student operator to fly independently.

The research group comprised 21 student operators, and the study consisted of 3 stages. The first stage was for each respondent to complete a personal questionnaire. The second stage was the referenced cardiovascular system work measurement, performed while the student operator was resting. The last stage was the measurements of cardiovascular system work taken during the training. The characteristics of the group of student operators will be discussed based on the results from the personal questionnaire. Among the participants, 95% do not have any aviation experience or qualifications. The remaining 5% (1 person) holds a Commercial Pilot License Aeroplane (CPL (A)). An important aspect of this research is the ability to operate the transmitter used to move the UAV which has structure and functions similar to controllers used in computer games, for example, role-playing-games (RPGs). The data collected are presented in Figure 1. Of all respondents, 62% play RPGs, and 38% of them do so with a frequency of 2 to 3 times a month. Additionally, 57% of the respondents use flight or driving simulators. In this case, the frequency is evenly divided, and among that 57% of respondents, 25% use them once a week, use them 2 to 3 times a month, use them less than 2 to 3 times a month or have used this form of computer game in the past. Additionally, hand–eye coordination and a person’s reaction speed are also important. Almost 60% of the student operators assess their reaction speed as good, 19% as very good, 19% as average and 5% as bad (Figure 1a). Furthermore, 52% of the respondents described their cognitive abilities (perception, attention, reasoning and memory) as good, 43% as very good and 5% as average (Figure 1b). This was a completely subjective assessment of reaction speed and cognitive abilities from the participants’ point of view. The largest number of candidates (85%) belonged to the age group of 24 and below, while the rest represented the following age groups: 35 to 39 years old (5%), 40 to 45 years old (5%) and 46 to 49 years old (5%) (Figure 1c). The questionnaire also took into account the motivation to obtain a certificate of qualification for a UAVO—57% were very interested in obtaining a certificate, 10% needed it to perform for their paid work and 33% did not care about obtaining a certificate (Figure 1d).

The study of student operators took place at Poznań–Kobylnica Airport (EPPK) during practical UAV training. Training lasted two days and covered 4 h of flights (according to [16]) in total. The exercises were performed using a helipad which was used for take-off and landing. Basic UAV manoeuvres were executed over the course of training.

The DJI Phantom UAV was used to perform practical training, while the POLAR H10 was chosen to carry out research into cardiovascular work. The POLAR H10′s sensor collects data about different parameters, such as heart rate and the length of RR intervals. This type of sensor (in belt form) was chosen for its comfort and ease of use. The POLAR belt is fully non-invasive and allows participants comfortable functionality while performing exercises. For each of the persons examined, the measurements featured a reference measurement, and the test was performed under the candidates’ resting conditions. While taking this measurement, the subject was in a seated position and was not distracted by anybody or any tasks. The second stage was the comparative measurement. During this, the student operators performed exercises related to the practical part of the training. The reference measurement lasted about 10 min, while the comparative measurement took 60 to 90 min.

3. Results

The research focused on cardiovascular system work—more specifically, on heart rate variability (HRV). The values of this factor can be an adequate physiological correlate of cognitive functioning. The HRV parameter is considered an indicator of autonomic control [20]. The most commonly used methods of quantitative and qualitative analysis of heart rate variability can be divided into several groups: time-domain methods, frequency-domain methods, time-frequency-domain methods and nonlinear methods. Based on the measurements performed, time-domain and nonlinear analysis were chosen to perform calculations. The measurement data used to determine the parameters that describe the HRV were analysed using dedicated software (Kubios HRV Standard 3.5.0 software). The first stage of the analysis was to remove artifacts from the measurements. After their removal, the record from the measurement did not present the direct length of RR intervals, but rather NN intervals, which are the values of RR intervals obtained after removing noise. Based on prepared samples, the parameters of the HRV were determined. HRV is expressed by differences in the intervals between the peaks of R waves (Figure 2), consecutive QRS complexes of sinus origin which describe the stimulation of the muscles and of the heart ventricles.

The samples used in the analysis were 5 min long, which is in line with the principles of conducting short-term variability (STV) analysis. The material was selected, and then, using the STATISTICA software (STATISTICA 13.3 software), the parameters for the time-domain and nonlinear analysis were determined. For a given subject, the parameters were obtained for the reference measurement (at rest) and the comparative measurement (during training).

The following variables were selected as part of the analysis:

• $\overline{x}$ RR—average length of RR interval (ms),
• $\overline{x}$ HR—average heart rate (bpm),
• HR max.—maximum value of heart rate (bpm),
• HR min.—minimum value of heart rate (bpm),
• SDNN—standard deviation of NN interval (ms),
• RMSSD—root mean square of successive differences between each RR interval,
• HRVI—histogram density function for the N interval divided by its height. The parameter is usually used for 24 h measurements,
• SD1 parameter—determined on the basis of the Poincarè diagram. This SD parameter is perpendicular to the line of identity. This indicator is characterised by a temporary variability (ms) value. An example of a Poincarè diagram for one of the subjects is shown in Figure 3,
• SD2 parameter—determined on the basis of the Poincarè diagram. This SD parameter is along the line of identity. This indicator characterises long-term heart rate variability and provides information about general variability,
• SD1/SD2—relationship between SD1 and SD2.

Figure 3. An example of a Poincarè plot for one of the student operators with the identity line marked and the parameters SD1 and SD2.

Figure 3. An example of a Poincarè plot for one of the student operators with the identity line marked and the parameters SD1 and SD2.

After analysis in KUBIOS, each respondent was described by 10 variables in 2 types of measurements (a total of 20 variables per participant).

Using the determined variables, statistical analysis was performed. This way, it was possible to perform tests for normality of distribution in individual variables, as well as significance tests. To check the normality of the distributions, the Shapiro–Wilk test was used. The results of the test were analysed with a significance level of 0.05 (

Table 1. Results of normality distribution tests and significance tests for individual variables (own elaboration). By * marked variables for reference measurements.

Variable	Shapiro-Wilk’s Test	p-Value	Relevance	t-Student Test	Wilcoxon’s Test	p-Value	Relevance
$\overline{x}$ RR	0.91301	0.06298	p > 0.05	0.40145	–	0.69234	p > 0.05
$\overline{x}$ RR *	0.96103	0.53731	p > 0.05
$\overline{x}$ HR	0.95959	0.50792	p > 0.05	1.66	–	0.11209	p > 0.05
$\overline{x}$ HR *	0.95766	0.47021	p > 0.05
SDNN	0.97111	0.75752	p > 0.05	2.81499	–	0.01070	p < 0.05
SDNN *	0.95476	0.41748	p > 0.05
RMSSD	0.72027	0.00005	p < 0.05	–	1.33817	0.18084	p > 0.05
RMSSD *	0.90712	0.04818	p < 0.05
HR max.	0.95688	0.45555	p > 0.05	3.08196	–	0.00588	p < 0.05
HR max. *	0.96687	0.66323	p > 0.05
HR min.	0.92975	0.13609	p > 0.05	8.30978	–	0.00045	p < 0.05
HR min. *	0.94175	0.23591	p > 0.05
HRVI	0.90828	0.05079	p > 0.05	1.78349	–	0.08969	p > 0.05
HRVI *	0.95707	0.45926	p > 0.05
SD1	0.72176	0.00005	p < 0.05	–	1.40769	0.04923	p < 0.05
SD1 *	0.90643	0.04671	p < 0.05
SD2	0.95271	0.38289	p > 0.05	6.23257	–	0.00194	p < 0.05
SD2 *	0.94891	0.32474	p > 0.05
SD1/SD2	0.51888	0.00000	p < 0.05	–	0.53874	0.59006	p > 0.05
SD1/SD2 *	0.61448	0.00000	p < 0.05

). As can be seen in Table 1, three variables were not normally distributed. Testing was necessary to properly select the type of significance tests. For variables with a normal distribution, a Student’s t-test of significance was used, and for variables that were not characterised by a normal distribution, the Wilcoxon test was used [32].

Table 1 shows the significance test results for each tested variable. At the significance level of 0.05, significant differences occurred between the reference and comparative measurement for five variables: SDNN, HR max, HR min and parameters SD1 and SD2 (marked in bold in Table 1).

Based on statistical analysis (Table 1), there are five designated variables that show significant differences and can characterise candidates’ psychophysical state. In order to facilitate the analysis and to identify possible relations between individual participants of the study, the research group was divided into smaller clusters. In each cluster, there are research objects that obtained similar values of the five analysed variables (parameters). According to this, the next stage of the calculations was to perform a multivariate analysis which aimed to show the groups formed among the participants.

The multivariate method Involves several different classification algorithms. As part of the investigation, group analysis was carried out to organise the data. The first stage of creating clusters is the selection of an appropriate algorithm, including the agglomeration method. The result of the first stage is a hierarchical tree presented in Figure 4b for the collected data.

The next step in creating clusters is to find out where significant change occurs in the distance between each created structure. Figure 3 shows the distance for individual steps in creating structures and the place (red line) where it was decided to divide the created structure into individual groups.

As a result of the division, five groups were obtained, and then their evaluation was carried out using the total figure index GSI, defined by formula 1 [21].

$$GSI=\frac{1}{g}{\sum }_l^{g}S\left(S_l\right)$$

(1)

where:

• GSI—total figure index,
• g—the number of groups resulting from analysis, sl—group l, with l = 1, 2, …, g,
• S(sl)—partitive total figure index designated for the group ${\mathrm{S}}_{\mathrm{l}}$

The total figure index (GSI) can take values in the range [−1, 1] to evaluate a single grouping case. The following interpretation is used for specific ranges of GSI values [18]:

• GSI > 0.70: strong structure of the groups obtained,
• 0.70 ≥ GSI> 0.50: correct grouping structure,
• 0.50 ≥ GSI> 0.25: weak grouping structure,
• GSI ≤ 0.25: no group in a given set.

The obtained value of the total body physique index GSI, which is equal to 0.55, indicates a correct grouping structure. This means that such prepared groups of respondents can be used in further data analysis.

Table 2. The values of variables characterising created groups and each tested student operator (own elaboration).

Group	Operator	SDNN	HR Max.	HR Min.	SD2	SD1	SD1/SD2
S1	6	29.371	147.42	67.345	40.082	10.911	0.272217
S2	14	48.059	104.6	60.546	65.414	18.428	0.281713
	16	55.334	111.61	64.879	75.487	20.633	0.273332
	18	56.767	104.09	64.483	75.81	26.252	0.346287
S3	1	34.9	101.84	64.949	47.687	12.835	0.269151
	5	31.562	123.56	61.158	43.761	8.845	0.202121
	9	38.394	110.74	63.83	51.012	18.639	0.365385
S4	3	26.392	124.48	80.994	35.558	11.248	0.316328
	4	24.385	122	85.33	48.56	8.353	0.172014
	7	28.666	134.53	82.192	39.789	7.8065	0.196197
	17	19.922	122.7	84.246	27.458	6.3306	0.230556
	21	21.04	114.72	83.025	28.85	7.2397	0.250943
S5	2	48	131.23	83.07	47.697	48.204	1.01063
	8	32.675	124.9	73.638	44.522	12.35	0.277391
	10	37.264	123.86	72.191	51.415	11.569	0.225012
	11	34.327	117.1	76.2	47.145	11.569	0.245392
	12	36.634	127.01	78.016	50.075	13.285	0.265302
	13	35.357	113.12	70.738	48.717	11.242	0.230761
	15	43.162	125.79	71.548	59.862	11.966	0.199893
	19	45.754	121.16	67.838	60.54	22.88	0.377932
	20	39.52	121.36	78.222	54.5	12.421	0.227908

shows the collected values of variables characterising the created groups and each tested student operator. The SDNN parameter is the standard deviation of the NN intervals—they are RR intervals without artifacts. According to the literature analysis [19,20,33], the smaller the deviations the examined person obtains, the greater the cognitive load of a given task. The normal value of the SDNN parameter for a healthy human, at resting conditions, is 50 ± 15 ms [18]. By analysing the results obtained from the measurements (Table 2) taken during the training exercises, it was shown on the basis of calculations that this parameter reached the value of 36 ± 10 ms during training. The next parameters analysed are SD1 and SD2. These deviations are descriptors determined on the basis of Poincarè charts (Figure 3). Both parameters correspond to standard deviations, and their definitions are included at the beginning of this chapter.

SD1 data provides information on short-term variability, and the SD2 parameter describes long-term variability. The SD1 parameter under resting conditions, in a healthy adult human, reaches a value of 66 ± 16 ms, while the SD2 parameter is 93 ± 19 ms [18].

Based on the analysis of the results obtained from the candidates during the practical training, it can be concluded that the values achieved for the parameters SD1 and SD2 were 15 ± 9 ms and 50 ± 13 ms, respectively. The ratio of the SD1/SD2 parameters characterises the equilibrium of the autonomic system, and its normal value is 0.29 ± 0.16. During the training exercises, respondents obtained values in the range of 0.7 ± 0.2. Additionally, an analysis of pulse parameters was performed during the training exercises. These are not parameters that characterise heart rate variability, but they showed significant differences from resting values in subjects—therefore, the decision was made to use them in the analysis. The literature analysis shows that the heart rate of a healthy adult human should be between 60 and 90 beats per minute (bpm) [18,20]. During the practical part of the training, the student operators obtained values in the range of 120 ± 11 bpm for the maximum value of HR max and 73 ± 8 bpm for the minimum value of HR min.

After analysis of the five significant parameters was performed, it was decided to elaborate the reference values of these parameters. This task was the main undertaking of this article. Elaboration was performed in the following steps:

• Means and standard deviations for each of the significant parameters in a given research group were calculated,
• The values of significant parameters obtained by each subject were analysed in terms of which candidates significantly “deviated” from the calculated averages, taking into account the deviation values. It was important to check the size of the differences (in subjects with “outliers”) compared with the calculated ranges,
• The calculated ranges ($\overline{x}$ ± SD) were compared with literature sources (based on studies related to the population) specifying what the intervals are for analysed parameters in healthy adults at rest. Thanks to these comparisons, it was possible to control the obtained results in terms of whether they are different from the range of normative values for healthy people. It turned out that the calculated ranges are smaller than or similar to the normative ranges in healthy adults at rest. This can provide information about cognitive load during training. It is assumed, based on other research [22,23,25], that an increase in cognitive load results in a decrease in HRV parameters. This influenced the reckoning that they are appropriate in comparison with the currently prevailing norms (based on studies related to the population) for the analysed parameters,
• The calculated values of averages and deviations were adopted as threshold values for the developed reference systems.

Based on the analysis of literature sources [22,23,25] and results obtained during this research (Table 2), reference values for the parameters of heart rate variability and the heart rate recorded by the student operators during training were summarised (

Table 3. Developed reference values of ranges for HRV parameters in UAVO training [13,14].

HRV Parameters	$\mathbf{STV} \mathbf{Norms} \overline{[\overline{\mathit{x}}}\left(\mathit{S}\mathit{D}\right)]$	$\mathbf{Reference} \mathbf{Values} \mathbf{for} \mathbf{UAVO} \overline{[\overline{\mathit{x}}}\left(\mathit{S}\mathit{D}\right)]$
SDNN	50 (16)	36.5 (10)
SD1	66.36 (16)	15 (9)
SD2	93.24 (19)	50 (13)
SD1/SD2	0.726 (0.06)	0.29 (0.16)
HR max.	-	120 (11)
HR min.	-	73 (8)

). In Table 2, the values marked in bold significantly differed from the developed ranges. As can be seen, most outliers’ values belong to structures 2 and 4.

Comparing the developed ranges to the measurement results obtained for each of the respondents (Figure 5, Figure 6 and Figure 7), it can be noticed that the candidates’ results from groups 2 and 4 differ from the established reference values. In the study, each of the subjects was analysed individually, and the research objects that deviated from the developed reference values were indicated. Such an approach was used due to the possibility of the reference ranges’ further use. The direction of further work will be to adapt the resulting intervals to, for example, a system for monitoring the psychophysical state of the candidate so that in the future it will be possible to carry out an objective assessment of his condition during training. For individual parameters characterising HRV, subjects who achieved values different from those in the developed ranges (Table 3) were selected. The results are shown in Figure 5, Figure 6 and Figure 7.

The HRV parameters obtained by the respondents from group 2 (subjects 14, 16 and 18 in Table 2) are greater than the adopted norms. This could indicate a light cognitive load. On the basis of the values obtained, it can be concluded that there was a slight activity of the sympathetic nervous system, which can prove a slight mobilisation of the organism (in the physiological way) to action [22,23,25,26,27,28,29,30,31]. The subjects in group 4 (subjects 17 and 21) obtained lower values for individual HRV parameters compared to the reference ranges. Apart from subjects in groups 2 and 4, significantly different values can be observed from the typical ranges for subjects 2 and 3. Student operator 2 achieved much higher values for the SD1 parameter and an SDNN that slightly exceeded typical values, while subject 3 obtained lower values in the case of parameter SD2. Therefore, it can be concluded that student operator 2 did not feel the cognitive load. Subject 3, on the other hand, achieved a similar tendency as in the cases of student operators from group 4. Smaller values prove that there may be a greater cognitive load on the individual respondents [20,22,24,26,27,28,29,30].

4. Discussion

Previous research has shown that based on HRV parameters, conclusions can be made on the psychophysical state of a subject [12]. This is possible due to the direct relationship between the parasympathetic and sympathetic nervous systems [26,27,28,29,30,31]. Currently, HRV analysis is often used in examining trainers/athletes. Based on such HRV analysis, training programs are created [34]. This parameter is also used in research on emotional control. Furthermore, one of the research publications regarding triathletes [35] shows that exercise intensity is the main factor in HRV reduction. Based on other research [22,23,25], it can be deduced that an increase in cognitive load results in an increase in heart rate and a decrease in HRV parameters (also confirmed through one of the performed research results). Most participants have smaller HRV parameter values during training than during rest, which confirms this dependence. The results from one of the studies demonstrated the possibility and benefits of in-flight HRV-based workload measurements using biometric sensors that can be integrated into the cockpit environment. The analysis performed in [36,37,38] clearly indicates that HRV analysis can be used to assess the task load of pilots, even in real conditions. The results from one of the studies indicate that, in general aviation, it is not possible to assess the pilot workload level just by using subjective assessment methods. The research describes the possibility and advantages of using HRV parameter measurement to analyse workload during flight operations and exploiting biometric sensors that can be easily implemented into the cockpit environment both in real and simulated flights [39]. Another study concerns direct UAV field tests involving the following hypothesis: mission difficulty in UAV operation affects both subjective and objective measurements of mental workload and also affects a number of failures in UAV operation. The results showed that mission difficulty affected the scores of all three assessment tools significantly. Mission difficulty also affected the number of failures and RR interval significantly [12]. The results obtained in the field of assessing the psychophysical state of operators and their cognitive load are comparable to the results obtained in the research included in this article. The research confirmed the thesis that HRV parameters are significantly different during performed tasks with cognitive load than they are at rest. In this research, values of HRV parameters during UAVO training were significantly different compared to the standard values of these parameters in a short-time measurement [17,40].

5. Conclusions

The conducted research together with the analysis allow for the formulation of the following conclusions:

• The parameters of HRV are sensitive to the functioning of the autonomic system, which allows for the analysis of the psychophysical state of the operator;
• In most of the respondents (80%), individual HRV parameters were lower during training tasks than at rest;
• It is possible to determine the values of reference ranges in HRV parameters that occur during the UAVO training—the main task of this article was achieved;
• The performed research confirms the information that appears in the literature: that with a higher cognitive load in task, HRV parameters will decrease. Subjects pointed out in analysis achieved smaller values of parameters than were found in the reference ranges, which means they experienced cognitive overload;
• In the cases of the student operators numbered 14, 18, 17 and 21, the analysis of their psychophysical state, using the parameters of heart rate variability, proved the presence of a significant or small cognitive load.

In the study, the decision was made to analyse each of the subjects individually and to indicate research objects that deviate from developed reference values due to the assessment of the possibility of the further use of the resulting ranges. The direction of further research will be to adapt the resulting intervals to, for example, a system for monitoring the psychophysical state of a candidate, so that in the future it will be possible to carry out an objective assessment of his condition during training. The developed reference values are the basis for developing a method that can be used as additional support for instructors, especially beginners, and make it easier for them to make an assessment. The research assumed that the proposed method (in case of further development) does not have to be implemented as a mandatory student operator assessment system. This is a proposal to improve the assessment of candidates and propose a solution that may have a positive impact on the safety of commercial UAV operations. The presented solution is a typical research approach to the development of UAVO training, not an implementation one.

The psychophysical state of a UAV operator is a meaningful aspect in terms of practical student training. The extension of training by using objective methods to perform psychophysical analysis (for example, HRV parameter measurements) as an element of the instruction process can result in the improved assessment of future operators. The conducted research and its analysis will be used to develop a proprietary method increasing the effectiveness of training in general aviation. They are the base for further considerations and the development of an algorithm supporting the assessment of the psychophysical condition of students before and during practical flight exercises.

Connecting HRV parameters with, for example, gaze behaviour [11] may lead to a complex psychophysical state assessment method. It may lead to the better identification of difficult exercises, which could result in greater UAVO training effectiveness. This type of method and experiment may also be important for other training areas, such as PPL(A) training. Further work will include expanding the research group and the research area to include training for other aviation qualifications. UAV training is the first area that was focused on in the presented research due to its dynamic development and the increasing demand for well-trained operators. Therefore, the addition of psychophysical state assessment to the training process can significantly reduce the risk of dangerous events occurring in UAV flights, caused by human error.