A
OBJECTIVELY EVALUATING SHORT-STROKE MOTION SYSTEMS INHELICOPTER FLIGHT SIMULATION
B.J.V. (Boris) Englebert1*and G.H.J. (Guido) Tillema2
1, 2 Aerospace Operations Training & Simulation Department, Royal Netherlands Aerospace Center, Amsterdam, 1059 CM, the Netherlands
*Corresponding author. E-mail: boris.englebert@nlr.nl;
Contributing author: guido.tillema@nlr.nl
Abstract
As helicopter flight simulation transitions toward eXtended Reality (XR) training devices, integrating short-stroke motion systems offers cost-effective, compact alternatives to traditional Full Flight Simulators (FFSs). However, the evaluation and tuning of such motion systems remain largely subjective, despite their importance in mitigating simulator sickness and enhancing training effectiveness. This paper introduces an objective evaluation framework to assess the performance and limitations of short-stroke motion systems in XR helicopter simulators. Using a servo-based 6 degree-of-freedom platform with a classical washout algorithm, the framework employs frequency-domain analysis via the Objective Motion Cueing Test (OMCT) and time-domain analysis of representative helicopter maneuvers. Results reveal that short-stroke systems are well-suited for cueing low-frequency, sustained accelerations and slow rotational movements, but struggle with high-frequency content due to motion envelope constraints. The study provides actionable insights into optimizing cueing strategies and informs recommendations on the effective use of short-stroke motion systems in helicopter flight simulation training.
Keywords:
flight simulation
motion cueing
short-stroke motion systems
objective motion cueing test (OMCT)
1.
Introduction
The helicopter flight simulation industry for flight crew training is currently undergoing a transformative shift, where eXtended Reality (XR) Flight Simulation Training Devices (FSTDs) are increasingly often used for flight crew training in lieu of traditional Full Flight Simulators (FFSs) [1, 2]. Such FSTDs that incorporate XR – an overarching term that covers concepts such as Virtual Reality (VR), Mixed Reality, and Augmented Reality – are characterized by the projection of the simulated environment through a Head Mounted Device (HMD), which allows the traditional visual display system and (part of) the cockpit instruments to be replaced by virtual equivalents [3, 4]. Unlike conventional FFSs, which require large dedicated spaces, substantial infrastructure, and high operational expenses, XR-based FSTDs provide a compact, modular alternative as a result. Consequently, XR FSTDs are associated with significant reductions in upfront capital investment, maintenance costs, and footprint costs, when compared with conventional FFSs. This, combined with the fact that the fidelity of XR FSTDs can be targeted to provide effective flight crew training, has been a driving factor behind the recent shift from flight crew training on FFSs to XR FSTDs [5–7]. In accordance with the definitions used by, e.g., the EASA CS-FSTD(H), an FFS features a motion system that provides key motion feedback and motion cues [8]. Typically, such a motion system used in an FFS is a six degree-of-freedom (DOF) hexapod, Stewart motion platform. Even though the contribution of motion cues in flight simulation to flight safety and transfer-of-training for flight crews is a debated topic, studies have demonstrated that motion plays a key role in training effectiveness for novice flight crew members when it comes to external disturbances and the control of vehicles with low dynamic stability such as helicopters [9]. In addition, motion cueing can play a key role in mitigating simulator sickness, for which the sensory conflict between visual inputs from the simulated environment and vestibular inputs is an widely known and recognized cause [10, 11]. Proper motion cueing can as such lead to vestibular sensory inputs that are more consistent with the visual sensory inputs, as a result of which the sensory conflict and, consequently, the severity of simulator sickness is reduced. Since simulator sickness can be an important reason for flight crew training to be terminated prematurely, the reduction of simulator sickness severity can aid in enhancing flight crew training effectiveness in FSTDs [10]. The traditional Stewart platform used in an FFS is a component that requires a significant amount of space and substantial maintenance efforts and costs. As such, adding a traditional Stewart platform to an XR FSTD would negate the aforementioned cost benefits for these devices. In order to reap the benefits of motion systems and motion cueing with regard to training effectiveness and the mitigation of simulator sickness – which is a particularly prevalent issue in XR FSTDs due to the strong sensory mismatch between the virtual (visual) environment and the vestibular inputs when sitting still ([12]) – a small-footprint motion system can be incorporated in XR FSTDs [13]. Such a motion system is characterized by actuators that have significantly reduced maximum excursions and as such a limited motion envelope compared to traditional Stewart platform actuators. Such a motion system is therefore also referred to in this research as a short-stroke motion system. As a result of the limited motion envelope of short-stroke motion systems, the accelerations and rotations that can be cued are restricted, which imposes additional challenges for providing realistic motion cueing that is conducive to an effective simulated training environment. In an effort to provide guidance on the quality of motion cueing on short-stroke motion systems, EASA has set out guidelines for the design and qualification of short-stroke motion systems in a VR-based Robinson R22 Beta II FNPT II and Airbus Helicopters H125 FTD III, in the form of a special conditions paper published in 2023 [13]. In these special conditions, EASA also reaffirms the importance of proper motion cueing from short-stroke motion systems for training effectiveness through pilot immersion and for the prevention of simulator sickness. Nevertheless, the guidelines and qualification standards proposed in these special conditions are, for a large extent, based on the qualification standards for traditional hexapod motion platforms in full flight simulators, and do not fully capture the unique characteristics of short-stroke motion systems and their applications in XR FSTDs for helicopter flight crew training. As a result, the design and evaluation of short-stroke motion systems and the tuning of the underlying motion cueing algorithm for helicopter XR FSTDs is primarily subjective in nature, and follows a similar approach as outlined in the existing EASA CS-FSTD(H) and FAA Part 60 qualification guidelines documents [8, 14]. Given the identified importance of motion cueing from short-stroke motion systems, and the current shortcomings in their evaluation, the effective deployment of short-stroke motion systems warrants a deeper and objective analysis to determine 1) what the limitations of these motion systems are, and 2) for which helicopter flight phases and manoeuvres they are suitable. This paper establishes an objective evaluation framework for the evaluation of short-stroke motion systems in helicopter flight simulation, and subsequently proposes recommendations for the use of short-stroke motion in (XR) helicopter flight simulation. In this framework, a short-stroke motion system with a classical washout motion cueing algorithm, as the industry-standard motion cueing algorithm, is subjected to both frequency domain and time domain-based analyses. The outcomes of these analyses can then be used to derive conclusions with regard to effective motion cueing strategies for short-stroke motion systems in helicopter flight simulation, as well as to the suitability of short-stroke motion platforms for various helicopter maneuvers and flight phases.
A
A
2. Background
To identify the applicable limitations and suitability of short-stroke motion systems in helicopter flight simulation, a thorough understanding of modern motion systems and how human motion perception can be leveraged to result in proper motion cueing is key.
2.1. Motion Systems and Motion Cueing
A motion system is defined as the collection of hardware and software components that translate digital inputs from the (helicopter) flight model to simulated motion from the motion system hardware. The motion system input, i.e., the output of the flight model, is the set of specific forces and rotational rates at the pilot position in the aircraft. This motion at the pilot position in the aircraft is then transformed to motion in the motion systems reference frame by the motion cueing algorithm. Subsequently, since the reference point about which the motion is computed by the motion cueing algorithm does not necessarily correspond with the pilot position in the simulator, a correction is applied to the computed motion, to the pilot head position at the Design Eye Reference Point (DERP). This correction is vital to prevent false cues, as the motion is sensed by the pilot at the DERP, as a result of which the motion experienced at the DERP should closely match the input motion. This motion is then sent to the motion system hardware, which then generates the simulated motion [15]. As the motion cueing algorithm is responsible for transforming the motion input (the flight model output) to motion system outputs, it plays an essential role in providing accurate motion sensations. These motion sensations are sensed by the pilot’s vestibular system, which is capable of perceiving specific forces – i.e. the combined acceleration from linear accelerations and accelerations resulting from gravity – and rotational rates, in six DOF [16]. The motion cueing algorithm therefore transforms the six DOF aircraft accelerations and rotational rates to consistent six DOF simulated motion, taking into account human perceptual thresholds and the constrained motion envelope of the motion system. The latter is of particular importance, since reaching the actuator limits at the boundaries of the motion envelope can result in false motion cues [15]. The most commonly used and industry-standard motion cueing algorithm to this day remains the classical washout algorithm (CWA). The CWA is a filter-based algorithms that splits the aircraft specific and rotational rates in three distinct channels, namely 1) the high-frequency translational channel that translates high-frequency aircraft specific forces into motion system linear accelerations, 2) the high-frequency rotational channel that translates high-frequency aircraft rotational rates into motion system rotational rates, and 3) the low-frequency translational channel that translates low-frequency aircraft specific forces to specific forces by means of simulator tilt, which is also referred to as tilt coordination [17]. The algorithmic structure of the CWA is visualized in Fig. 1, which also shows the distinct channels. In the two high-frequency channels – the translational channel and the rotational channel – the respective specific forces and angular rates are first scaled with fixed gains with the purpose of reducing the magnitude of both motion signals. These scaled motion signals are then fed into a high-pass filter, that serves to filter out low-frequency motion components that typically result in large motion excursions that would violate the motion envelope of the motion system. These high-pass filters also ‘wash-out’ the high-frequency motion components and ensure that the motion system returns to its neutral position [17]. As mentioned, the cueing of the low-frequency specific force components is instead done by means of tilt coordination, i.e., tilting the simulator to simulate sustained low-frequency accelerations by realigning the gravity vector in the DERP reference frame. Since the human vestibular organ senses specific forces, and cannot distinguish between linear accelerations and accelerations resulting from gravity, this provides the sensation of a sustained (i.e., low-frequency) acceleration. As can be inferred from Fig. 1, this is done by passing the scaled specific force signal through a low-pass filter in the tilt coordination channel. The resulting tilt angles are then superimposed with the angular motion resulting from the high-frequency rotational channel [17].
Fig. 1
Schematic of the CWA
(adapted from [15] and [18]
2.2. Motion System Evaluation
A
A
As mentioned, the motion system covers the software and hardware components than transform the simulated aircraft motion at the pilot position in the aircraft, to simulated motion at the pilot position (DERP) in the simulator. The dynamic behavior of the motion system is affected by three key components: the motion cueing algorithm, the dynamics of the motion hardware, and inherent digital delays within the system [15]. The most common qualification standards and guidelines for FSTDs, including the EASA CS-FSTD(H) and the Royal Aeronautical Society guidelines, stipulate hardware-based tests for the motion system, for instance with regard to leg balance and turnaround bumps. However, the evaluation and tuning of the motion cueing itself, is not covered as part of these tests, and remains subjective in nature in these standards and guidelines [8, 19]. As motion system hardware continues to mature and improve, the simulated motion fidelity becomes increasingly dominated by the quality of the underlying motion cueing algorithm, as a result of which it can be argued that the relevance of the hardware-based tests outlined in common qualification standards and guidelines with regard to the simulated motion fidelity diminishes [20]. A more holistic approach for the evaluation of motion systems was proposed by Advani & Hosman in 2006 in the form of the Objective Motion Cueing Test (OMCT) [21]. The OMCT measures the motion system frequency domain response, including both the motion hardware and software components, in all individual motion system axes for a set of sinusoidal inputs with predetermined amplitudes and frequencies. The motion system response in the OMCT is consequently expressed as the linear transfer function from the flight model output (i.e., the motion at the pilot position in the aircraft) to the motion system output (i.e., the motion at the DERP in the simulator), and captures the contribution of all three aforementioned components that shape the dynamic behavior of the motion system response [15, 17, 22]. The OMCT was included by the ICAO in Doc. 9625, where its procedure with regard to the execution of the test and interpretation of the results can be found [18]. As described in ICAO Doc. 9625, the motion system frequency responses in the OMCT are computed for all direct specific force and rotational rate relationships, as well as for a set of spurious cross-couplings that arise due to the motion system hardware and motion cueing algorithm. In total, the OMCT features ten distinct frequency response function that define the motion system response ([18, 19]), which are described below. For a concise summary of the OMCT tests, see also the test matrix in Table 4.
1.
Motion system sway specific force response due to aircraft sway specific force input;
2.
Motion system roll rate response due to aircraft sway specific force input;
3.
Motion system surge specific force response due to aircraft surge specific force input;
4.
Motion system pitch rate response due to aircraft surge specific force input;
5.
Motion system heave specific force response due to aircraft heave specific force input;
6.
Motion system roll rate response due to aircraft roll rate input;
7.
Motion system sway specific force response due to aircraft roll rate input;
8.
Motion system pitch rate response due to aircraft pitch rate input;
9.
Motion system surge specific force response due to aircraft pitch rate input;
10.
Motion system yaw rate response due to aircraft yaw rate input;
In the overview above, the direct input-output frequency response functions are computed in tests 6, 8, and 10 for the rotational rates, and in tests 1, 3, and 5 for the specific forces, respectively. Tests 7 and 9 determine the cross-coupling effects for surge and sway motion system responses to pitch and roll inputs, which may, e.g., be indicative of false motion cues that occur when the correction to the pilot position (DERP) reference frame is not done correctly. Finally, tests 2 and 4 give insight into the cross-coupling effects for pitch and roll motion system responses to surge and sway inputs, which can be associated to false motion cues arising from an improperly tuned tilt coordination channel in the CWA [18, 19]. The frequency responses for all OMCT test cases are computed based on twelve sinusoidal inputs with a frequency that ranges from 0.10 rad/s to 15.8 rad/s, which was deemed representative of common aircraft motion [21]. The amplitudes for the sinusoidal inputs are, in most cases, 1.0 m/s2 and 1 deg/s for the specific forces and rotational rates, respectively. The amplitudes for the rotational rate sinusoidal inputs are reduced towards the low and high end of the indicated frequency range, in an effort to minimize non-linear effects in the motion system response at these frequencies [18, 20]. An overview of the OMCT sinusoid frequencies and amplitudes can be found in Table 4. It is important to note that the aim of the OMCT is to predominantly measure the linear motion system response, while minimizing non-linear contributions [15]. Since both helicopter motion and the underlying motion cueing algorithm can have significant non-linearities (such as rate limiters in the motion cueing algorithm), however, the OMCT can give a limited insight into the simulated motion fidelity and applicability for the particular aircraft type and FSTD training needs [20]. In order to acquire a more complete assessment of the motion system response and simulated motion fidelity, the objective evaluation of the motion system can, e.g., be augmented with the (time domain) examination of the motion system response for a set of representative helicopter maneuvers, that capture non-linearities that are not covered by the OMCT.
A
Fig. 2
The short-stroke motion system as part of the MoVR-X simulator at the Royal NLR
2.3. Short-Stroke Motion Systems
Similar to traditional Stewart platforms, the evaluation and tuning of short-stroke motion systems is primarily subjective. Whilst several traditional Stewart platforms have been subjected to the OMCT (for an overview, see [15]), no concrete and publicly available objective evaluation data for short-stroke motion systems is obtainable. This, combined with the fact that the qualification guidelines for short-stroke motion systems published in the EASA special conditions are directly based on the standards for traditional Stewart platforms in an FFS Level B ([13]), reveals a gap in the objective evaluation of the short-stroke motion systems in particular. This gap is particularly relevant considering the inherent hardware differences between short-stroke motion systems and traditional Stewart platforms – not only in terms of a reduced motion envelope, but also in terms of the actuator mechanism, where short-stroke motion system often feature servo-based actuators as opposed to linear actuators.
As such, this paper describes an objective evaluation framework and corresponding analysis of a short-stroke motion system using the OMCT as a frequency domain motion system evaluation approach, supplemented with a time domain analysis of several relevant helicopter maneuvers. The precise experimental setup, objective evaluation framework, and test approach, are discussed next.
3. Method
In order to apply the objective evaluation framework, a helicopter FSTD setup that features a short-stroke motion system at the Royal Netherlands Aerospace Center (NLR) was used. The details of this setup, as well as the precise procedures for the frequency domain analysis, featuring the OMCT, and the time domain analysis are discussed in this section.
3.1. Short-Stroke Motion System and Helicopter Flight Simulator Setup
The measured DERP positions and orientations are passed through a noise filter, being a low-pass filter with an appropriately selected break frequency to retain the OMCT signal information, to filter out sensor noise. After this, these signals are differentiated using a central differences differentiation scheme to attain the corresponding accelerations and rotational rates. The specific forces
and
for the respective surge and sway axes are reconstructed by combining the linear accelerations
and
at the DERP with the pitch
and roll
angles, respectively. This is done in accordance with Eq. 1 and Eq. 2, respectively, and a right-handed body reference frame with the
-axis pointing down, to incorporate the effects of tilt coordination.
Since numeric differentiation of discrete time series data can introduce spurious noise, the differentiated signal measurements are passed through a smoothing filter to properly reconstruct the specific forces and rotational rates at the DERP.
3.2. Frequency Domain Motion System Analysis
A
For the objective frequency domain analysis of the motion cueing, the OMCT is applied. The OMCT test cases outlined in Section 3 and described in [18] are summarized in Table 3, while the frequencies and amplitudes for the applicable sinusoidal inputs can be found in Table 4.
Since the influence of the short-stroke motion system limitations on the motion response is of particular interest, it was decided to generate OMCT results for both the isolated motion cueing algorithm output, as well as for the complete motion system response. The simultaneous analysis of these two output signals will reveal which potential limitations in the motion system response are generated by the motion cueing algorithm, and which limitations can be attributed to the short-stroke motion system hardware.
3.3. Time Domain Motion System Analysis
As mentioned previously, the aim of the OMCT is to primarily measure the linear motion system response per motion system axis, while omitting non-linear effects as much as possible [15]. Since only examining the linear motion system response can paint a limited picture of the motion system response fidelity and suitability for the FSTD training needs, it can be useful to extend the evaluation of the motion system beyond the OMCT. This is of particular relevance for servo-based short-stroke motion systems used in helicopter FSTDs, as non-linearities in the motion system response can arise from 1) cross-couplings in the motion system hardware due to the servo-based actuators, and 2) inherent non-linearities in helicopter flight characteristics. As such, it was decided to expand the objective evaluation framework by means of time domain analyses of the short-stroke motion system response for several representative helicopter maneuvers. To this end, a Royal Netherlands Air Force pilot flew a set of Mission Task Elements (MTEs) from the ADS-33E-PRF ([23]), which is a US Army standard for the evaluation of the handling qualities of rotary wing aircraft, in the MoVR-X simulator. In this setup, the MoVR-X simulator was equipped with COTS helicopter flight controls, a Varjo XR-3 HMD, and a high-fidelity UH-60 Black Hawk flight model The Royal Netherlands Air Force pilot was a 38-year-old male with an active military helicopter pilot license for the CH-47F Chinook, on which he had accumulated 2480 flight hours at the time of the execution of the MTEs in the MoVR-X simulator. The MTEs were selected on the premise that they together should 1) excite all motion system axes, and 2) cover a span of frequencies and amplitudes for translational and rotational motion. An overview of the MTEs used for the time domain motion system analysis is given below, including relevant characteristics with regard to the applicable motion system axes.
Depart-Abort MTE
In the depart-abort MTE, a longitudinal acceleration and subsequent deceleration is executed along a straight 800-ft track. After starting the maneuver from a stable hover, the acceleration shall be initiated by means of a change in pitch attitude, followed by a change in pitch attitude to start the deceleration. The maximum changes in pitch attitude shall be limited to 20 degrees, while the acceleration and deceleration shall be executed in one smooth maneuver in such a way that a stable hover is reached at the finish line at the end of the 800-ft track. Motion system axes: Surge and pitch, due to the longitudinal acceleration and deceleration, and the pitch rate associated to changes in pitch attitude.
Slalom MTE
The slalom MTE comprises a set of slalom turns through a series of poles at 500-ft intervals, while maintaining a groundspeed of more than 40 knots and remaining below a radio altitude of 100 ft. As such, the motion characteristics related to the slalom primarily consist of lateral accelerations and changes in roll attitude. Motion system axes: Sway, roll and yaw, due to the lateral accelerations and the roll rate associated to lateral motion, as well as the yaw rate as a result of the required heading corrections.
Hover MTE
The aim of the hover MTE is to maintain a stable hover position for 30 seconds, by aligning two reference symbols spaced 75 ft longitudinally apart from each other in front of the hover position. The hover MTE therefore requires adjustments in primarily surge, sway, and heave, in order to maintain the proper hover position. Since these adjustments consist of small inputs on the translational axes, there is less risk of the limited motion system envelope being violated. Hence, it can be argued that a short-stroke motion system is particularly well-suited for such an MTE (or maneuver, more generally). Motion system axes: Surge, sway, and heave due to the required adjustments in longitudinal, lateral, and vertical position necessary to maintain a stable hover position in relation to the reference symbols.
In the time domain analysis of the motion system response for the MTEs described above, a distinction is made between the linear acceleration at the DERP from the translational channel of the CWA, and the specific force at the DERP due to the added effect of tilt coordination from the tilt coordination channel. This allows for the separate analysis of the motion system’s ability to provide high-frequency motion cues from direct displacement and resulting linear accelerations, as well as low-frequency, sustained motion cues from tilt coordination.
4. Results
As described in Section 4, the objective evaluation framework applied to the short-stroke motion system as part of the MoVR-X simulator consists of two distinct analyses, namely the frequency domain analysis in the form of the OMCT, and the time domain analysis by means of the helicopter MTEs. This section presents the results for both these analyses, as part of the objective evaluation framework.
4.1. Frequency Domain Motion System Analysis
Since the surge and pitch, as well as the sway and roll, axes are known to be coupled, the OMCT results for the excitation of these axes are presented a in tandem. As the heave and yaw axes are independent of any of the other motion system axes, they are analyzed separately.
Surge and Pitch Axes
A
Fig. 3
OMCT test case 3 (surge-surge) motion cueing and motion system response (see Table 4)
Fig. 5
OMCT test case 4 (surge – pitch) motion cueing and motion system response (see Table 4)
A
Figure 4 OMCT test case 8 (pitch – pitch) motion cueing and motion system response (see Table 4)This can be explained by noting that the tilting of the motion system from tilt coordination at lower frequencies inherently introduces a pitch rate, which can therefore result in a false pitch rate cue. Furthermore, the phase behavior is Fig. 5 is consistent with a signal that is attenuated by a first-order low-pass filter and subsequently passed through a differentiator that introduces a 90-degree phase lead, as it is necessary to differentiate the pitch angles computed in the tilt coordination channel to pitch rates for the results of this particular OMCT case. The results for OMCT case 9 (i.e., the surge cross-coupling response to a pitch rate input) are characterized by low gain scaling and consistent phase behavior between the motion system response and motion cueing response, implying that no significant cross-coupling effects can be noted in the motion system response. The Bode plots for this cross-coupling OMCT test cases are omitted for brevity.
Sway and Roll Axes
A
Analogous to the OMCT test results for the surge and pitch axes, the OMCT test results for the direct input-output responses for the sway and roll axes (OMCT tests 1 and 6), are visualized in Fig. 6 and Fig. 7, respectively. The OMCT results for the direct sway and roll input-output responses generally show the same trends as the surge and pitch axes OMCT results, where similar low-pass filter and high-pass filter behavior is visible in terms of the gain and phase shifts over the OMCT frequency range. The differences between the motion system response and the motion cueing response are also comparable to what was found for the surge and pitch axes results. Specifically, for the sway axis in Fig. 6 it can be seen that the motion system response follows the motion cueing response (and therefore the commanded motion) closely at the lower frequencies, both in terms of gain and phase. It is also clear that the motion system response deteriorates with respect to the motion cueing response at the higher frequencies, evident from a lower gain scaling an additional phase lag. As for the roll rate, Fig. 7 demonstrates that the motion system response exhibits similar phase and gain behavior as the motion cueing response at the lower frequencies, while a sharp decrease in the gain scaling and differing phase behavior occurs for the motion system response at the higher frequencies, with respect to the motion cueing response. Also similar to the surge and pitch axes OMCT test results, the cross-coupling OMCT test results (test cases 2 and 7) for the sway and roll axes show a noticeable cross-coupling effect for OMCT case 2, which measures the roll rate response to a sway input. Shown in Fig. 8, the results for OMCT case 2 show a gain scaling in the lower frequency range for the roll rate, similar to what occurs for the surge-pitch cross-coupling results for OMCT case 4 in Fig. 5. This implies that a similar effect is present in the sway and roll axis when the tilt coordination mechanism is active at lower frequencies, where the introduction of tilt results in an inherent roll rate. The results for OMCT case 7 (i.e., the sway cross-coupling response to a roll rate input) exhibit low gain scaling across the entire frequency range, again indicating that the cross-coupling effects in the motion system response are minimal. The Bode plots for this OMCT case are not depicted for brevity.Fig. 8
OMCT test case 2 (sway – roll) motion cueing and motion system response (see Table 4)
Fig. 6
OMCT test case 1 (sway – sway) motion cueing and motion system response (see Table 4)
Figure 7 OMCT test case 6 (roll – roll) motion cueing and motion system response (see Table 4)
Heave Axis
Fig. 9
OMCT test case 5 (heave – heave) motion cueing and motion system response (see Table 4)
Following the CWA structure in Fig. 1, the heave axis is the only translational axis in which the signal is passed through only a high-pass filter. The associated effects of this high-pass filter are clearly visible in the OMCT test results for the direct heave input-output responses, shown in Fig. 9, where the low-frequency sinusoids are attenuated, and only high-frequency responses show noticeable gain scaling. Consistent with the results for the surge and sway axes, it can also be observed that the motion system response gain and phase deteriorate compared to the motion cueing response for the higher frequencies, which is indicative of a poor motion system response compared to the commanded motion from the motion cueing algorithm at these frequencies. The phase behavior for the responses is consistent with a high-pass filter, which introduces a 180-degree phase lead at the lower frequencies that decreases as the input signal frequency increases.
Fig. 10
OMCT test case 10 (yaw – yaw) motion cueing and motion system response (see Table 4)
Yaw Axis
The OMCT direct yaw input-output test results (OMCT test case 10) in Fig. 10 are comparable to the OMCT results for the direct pitch and roll input-output test results, when it comes to the gain and phase behavior for a first-order high-pass filter. In addition, Fig. 8 demonstrates that the gain and phase decay at the higher frequency range for the motion system response relative to the motion cueing response is consistent to what can be observed for the other rotational axes as well.
4.2. Time Domain Motion System Analysis
As discussed before, the time domain motion system analysis results are structured by means of the various MTEs and the associated motion system axes described in section 4.3. The results presented in this section therefore consist of time traces of the MTEs for 1) the flight model output, as input to the motion cueing algorithm, and 2) the complete motion system response. Furthermore, a distinction is made between the linear acceleration measured at the DERP, and the specific force measured at the DERP for the translational axes that feature tilt coordination. This is done to identify the contribution of the linear accelerations, resulting from the direct movement of the motion system actuators within the limited motion envelope, to the overall specific force that can be sensed at the DERP. Additionally, this allows for a more detailed analysis of the short-stroke motion system’s suitability for providing high-fidelity motion cues at the DERP for both low-frequency and high-frequency content in the various MTEs.
Depart-Abort MTE (Surge and Pitch Axes)
A
The time traces for the surge linear acceleration and specific force as well as the pitch rate for the depart-abort MTE, measured at the DERP, are shown in Fig. 11 and Fig. 12, respectively. From Fig. 11, it is apparent that the contribution of the linear acceleration to the overall specific force measured at the DERP is minimal. This implies that the vast majority of the specific force originates from the tilt coordination to provide the low-frequency motion sensations associated with the acceleration and deceleration phases of the maneuver. Additionally, the higher-frequency motion ‘peaks’ in the flight model output in Fig. 11, which should primarily be cued through direct accelerations as a result of the high-pass filtering, are not visible in the measured linear accelerations at the DERP, which indicates the limited capability of the motion system to provide such cues. The measured pitch rate for the depart-abort MTE in Fig. 12 initially does not follow the flight model output consistently. Arguably, these ‘false motion cues’ in the beginning of the maneuver are a direct result of the motion system tilting to provide the corresponding specific forces, as they coincide with the start and end of the acceleration phase. Since the pitch rate does follow the general trend of the flight model output during the later stages (and majority) of the maneuver, albeit smaller in magnitude, it can therefore be argued that the motion system pitch rate response is capable of providing realistic motion cues for the depart-abort MTE.Fig. 12
Depart-abort MTE pitch rate time traces
Figure 11 Depart-abort MTE surge linear acceleration and specific force time traces
Slalom MTE (Sway, Roll, and Yaw Axes)
Figure 13, Fig. 14, and Fig. 15 show the time traces for the measured DERP sway linear acceleration and specific force, roll rate, and yaw rate, respectively, for the slalom MTE. Again, the sway axis motion system response exhibits noticeable similarities to the surge axis response, as Fig. 13 demonstrates that the contribution of the measured linear acceleration at the DERP is negligible compared to the overall measured specific force at the DERP. The implication of this observation is that, also in the sway axis, the measured specific forces are primarily provided through tilt coordination in response to low-frequency motion components. Contrary to the measured pitch rate response for the depart-abort MTE, however, the measured roll rate response for the slalom MTE shows a poor tracking of the flight model output, i.e., the input signal to the motion cueing algorithm. It can be argued that the underlying cause of this is twofold. Firstly, the flight model roll rate output features more high-frequency motion content (or in other words, greater rotational accelerations), than the flight model pitch rate output in Fig. 12. Since the mechanical links and actuators of the motion system are unable to provide these high rotational accelerations in the roll axis, and thus in essence collectively function as a mechanical low-pass filter, the motion system response is unable to accurately track the roll rate input signal. Secondly, false roll rate cues coincide with changes in the sign of the sway specific force, such as around
= 27 s in Fig. 13 and Fig. 14. Provided that the sway specific force originates from tilt coordination, it can be asserted that these false roll rate cues are direct result of the need to tilt the motion system for providing the correct specific force sensation in the sway axis. This is analogous to what was inferred for the interaction between the surge and pitch axes for the depart-abort MTE from Fig. 11 and Fig. 12. Finally, the yaw rate measured at the DERP for the slalom MTE, visualized in Fig. 15, exhibits similarities with both the measured pitch rate and roll rate for the depart-abort MTE and slalom MTE. That is, the measured yaw rate appears to track low-frequency motion content in the flight model output relatively well (again, albeit with a smaller magnitude), while high-frequency content from the flight model output signal seems to be largely absent in the measured yaw rate. In contrast with the pitch rate and roll rate, the yaw rate does not suffer from false motion cues as a result of couplings with other motion system axes, as the yaw axis is independent of the other axes, as a result of which false yaw rate cues are noticeably less frequent in Fig. 15, compared to the measured pitch rate and roll rate in Fig. 12 and Fig. 14, respectively.
Fig. 15
Slalom MTE yaw rate time traces
Fig. 14
Slalom MTE roll rate time traces
Fig. 13
Slalom MTE sway linear acceleration and specific force time traces
Hover MTE (Surge, Sway, and Heave Axes)
A
Contrary to the depart-abort and slalom MTEs, the hover MTE is characterized by smaller translational specific forces, as the aim of the hover MTE is to maintain a stable position for which occasional translational adjustments are necessary. The time traces for the surge, sway, and heave flight model outputs and motion system responses are shown in Fig. 16, Fig. 17, and Fig. 18. Logically, the flight model output specific forces for the hover MTE are noticeably smaller than in the depart-abort MTE and the slalom MTE As a result, it is striking that the motion system response specific forces in particularly the surge and sway axis in Fig. 16 and Fig. 17 exhibit a much closer match to the flight model outputs than in the depart-abort MTE and slalom MTE. Also for the hover MTE, it can be seen that the contribution of the linear acceleration to the overall specific force in the surge and sway axis either not substantial, or even constitutes a false cue (possibly still due to reaching the limits of the motion envelope). As such, it can be argued that the motion system response specific forces for the surge and sway axes are primarily generated through tilt coordination of the low-frequency flight model output content. Finally, it can also be observed from Fig. 17 and Fig. 18 that the maximum specific forces that are provided by the motion system in the surge and sway axes are between 0.5 m/s2 and 1.0 m/s2, which displays consistency with the maximum surge and sway specific forces in the depart-abort and sway MTEs. For the specific force response in the heave axis in Fig. 18, for which the specific force only consists of the linear acceleration provided by the motion system, it is apparent that the motion system response follows the general trend of the flight model output, albeit with a smaller magnitude compared to the surge and sway axes. Again, this is logical because of the fact that tilt coordination (and the associated effect on the cueing of low-frequency specific forces) does not occur in the heave axis, and because the gain in the filter parameters for the heave axis in the CWA is smaller than for the surge and sway axes (see also Table 2).Fig. 16
Hover MTE surge linear acceleration and specific force time traces
Fig. 17
Hover MTE sway linear acceleration and specific force time traces
Fig. 18
Hover MTE heave specific force time traces
5. Discussion
The aim of this research is to establish an objective evaluation framework for short-stroke motion systems in helicopter flight simulation, with the purpose of 1) identifying the limitations of such short-stroke motion systems, and 2) to make recommendations for which helicopter flight maneuvers these short-stroke motion systems are suitable. The objective evaluation framework consists of two evaluation approaches, one in the frequency domain, and one in the time domain. The frequency domain motion system analysis consists of the OMCT and focuses on the linear frequency responses of the motion system in response to sinusoidal excitations on the individual motion system axes over a range of characteristics frequencies [15]. The OMCT only gives limited insight into the fidelity of the short-stroke motion system for providing realistic motion sensations for helicopter maneuvers, which is why the evaluation is augmented by means of a time domain analysis. The time domain motion system analysis consists of the evaluation of the motion system response time traces for a set of MTEs from the ADS-33E-PRF ([23]) that covers all motion system axes. For the frequency domain motion system analysis, the frequency responses were shown in Bode plots. Most importantly, these Bode plots showed that the fidelity of the motion system response is relatively high for the lower frequencies in the OMCT frequencies range, apparent from a relatively accurate gain scaling and appropriate phase behavior, while the fidelity of the motion system response noticeably deteriorates for the higher frequencies, approximately above 2.0–3.0 rad/s. While this is true for the direct input-output frequency responses of all motion system axes, it is striking that the fidelity of the translational channels appears to be higher than the rotational channels, evident from a better accuracy of the motion system response gain. In the low-frequency range where this is applicable, the surge and sway specific forces are generated through tilt coordination, i.e., tilting the platform to realign the gravity factor in the DERP reference frame to artificially provide a sense of acceleration. Since this tilting is much less dynamic in nature than the linear accelerations generated by direct movement of the motion system actuators at higher frequencies, there is 1) a lower risk of violating the limited motion envelope, and 2) it is possible to sustain the specific forces from tilt coordination. As such, these findings are an indication that the short-stroke motion system is especially capable of providing sustained, low-frequency specific force sensations, such as sustained accelerations during a transition in terms of helicopter flight. This result is also echoed by the time domain motion system analysis results for the surge and sway axes. Specifically, the motion system response time traces for these axes all show that the contribution of the linear acceleration to the overall specific force is limited, from which it can be deduced that the majority of the generated specific forces for these axes originates from tilt coordination for the slower (i.e., low-frequency) motion components. For the faster (i.e., high-frequency) motion components in the surge and sway axes, it is clear from the time domain motion system analysis that the replication of the specific forces is either very limited, or results in false motion cues, presumably due to violations of the motion envelope. This is consistent with the OMCT results where, again, the gain scaling and phase behavior at these higher frequencies shows inaccuracies. Similarly, the time traces for the heave motion system response in the hover MTE, which does not feature tilt coordination, demonstrate that the motion system is capable of replicating the lower frequency trends, albeit with a much smaller magnitude, while the higher frequency motion components appear to be largely absent in the motion system response. This also shows consistency with the OMCT results for the heave axis, which demonstrated a relatively low gain scaling that rapidly deteriorates at the higher frequencies. In practice, this therefore implies that the motion system is capable of providing short-lived, low-frequency, low-acceleration cues in the heave axis. As for the time domain motion system analysis of the rotational rates, the general conclusions are arguably similar to those made for the heave axis. That is, the results for the pitch rate in the depart-abort MTE, and roll and yaw rate in the slalom MTE, demonstrate that slower rotational rates (i.e., low rotational accelerations) are replicated by the motion system with relatively high fidelity in terms of timing, while being smaller in magnitude. Higher-frequency rotational rates (i.e., high rotational accelerations) do not seem to be replicated by the motion system, and can mostly not be observed in the motion system response. The consistency with the OMCT test results for the direct input-output frequency response functions of the rotational axes is applicable here as well, where, as mentioned before, the frequency responses show a relatively high fidelity at the lower-frequency ranges that becomes gradually worse towards the higher end of the frequency spectrum. One caveat for the motion system response for the rotational rates, is the occurrence of false rotational cueing in conjunction with introducing tilt coordination for the cueing of sustained translation specific forces. To elaborate, the time traces for the depart-abort MTE and slalom MTE show false ‘spikes’ in the pitch and roll rate that occur simultaneously with changes in the sign of the respective (sustained) surge and sway specific forces, that do not occur in the flight model output that is passed to the motion cueing algorithm. These findings are corroborated by the frequency responses for the surge-pitch and sway-roll cross-coupling OMCT cases, or OMCT cases 4 and 2, respectively. That is, the frequency responses for these cross-coupling test cases show a non-negligible gain scaling at the lower frequencies, where the tilt coordination mechanism is active. The implication of this is that false rotational cues may be experienced upon the tilting of the motion system to provide longitudinal or lateral specific force sensations. The cause of this can be found in the parameters of the CWA, and in particular in the interaction between the settings of the low-pass filter and the rate limiter in the tilt coordination channel. Particularly, since the results demonstrated that short-stroke motion systems can primarily generate specific force sensations for sustained accelerations through tilt coordination, it may seem like an obvious strategy to increase the low-pass filter break frequency in the tilt coordination channel to allow for more (high-frequency) motion to be passed through this channel. This, however, also requires the motion system to tilt faster in order to accommodate changes in the specific force, which naturally introduces higher associated rotational rates. These rotational rates can be dampened by the rate limiter, which therefore plays a key role in preventing such false rotational cues. Thus, for the design of motion cueing algorithms for short-stroke motion systems, there appears to be a trade-off between 1) changing the CWA parameters to introduce more tilt coordination in an effort to cue more translational motion, and 2) to prevent false motion cues in the process as result of, e.g., inadequately modifying the low-pass filter break frequencies or rate limiter characteristics. It should be noted that it is to be expected that this phenomenon and associated effects are not fully captured in the OMCT results, as this constitutes non-linear behavior that is not contained in the linear frequency response functions generated in the OMCT. Reflecting further on the OMCT test results, it is striking that the OMCT frequency responses for the short-stroke motion system generally fall outside of the ‘fidelity’ criteria boundaries proposed by Hosman and Advani [15]. It should be noted, however, that these criteria boundaries were derived based on the OMCT test results for traditional Stewart platforms. As a result, it may not be realistic to assume that these criteria boundaries are achievable for short-stroke motion systems, for which the motion envelope is simply much more limited than for traditional Stewart platforms.
6. Conclusion
As the helicopter flight simulation industry increasingly transitions from traditional FFSs to XR-based FSTDs for flight crew training, the incorporation of short-stroke motion systems in XR-based FSTDs is rapidly becoming an important consideration. Due to the inherent differences between short-stroke motion systems and the traditional Stewart platforms used in FFSs, it is not self-evident that existing motion system design principles and qualification guidelines are applicable to short-stroke motion systems as well. This, in combination with the fact that the design and evaluation of these short-stroke motion systems is almost exclusively subjective in nature, demands a more thorough review and evaluation approach for such motion systems. This paper establishes an objective evaluation framework for short-stroke motion systems in helicopter flight simulation. Applied to a servo-based short-stroke motion system, the objective evaluation framework demonstrates that short-stroke motion system are particularly capable of providing motion sensations for the following motion characteristics and associated helicopter flight phases:
1
Sustained, low-frequency specific forces in the surge and sway axes. In terms of helicopter maneuvers, this could constitute a sustained longitudinal acceleration as part of a transition, or a sustained lateral acceleration as part of a circuit or pirouette maneuver at a constant lateral groundspeed. Importantly, the trade-off between the low-pass filter and rate limiter parameters in the classical washout algorithm should be considered to prevent false rotational rate cues.
2
Slow rotational rates, in all rotational axes. For helicopter flight, this could, e.g., be roll attitude and heading adjustments in a slalom maneuver, or pitch adjustments when performing a descent.
3
Short-lived, low-magnitude specific forces in the translational axes. In the heave axis in particular, the results show that the motion system can provide short-lived acceleration sensations due to direct displacement of the actuators, as long as these remain below a particular frequency (e.g., 2.0–3.0 rad/s for the motion system in this research).This cannot be sustained long, due to the limited motion envelope, however. For the use of short-stroke motion systems in helicopter flight simulation and training, this implies that it would, e.g., be possible to cue the short-lived vertical acceleration associated with the ‘landing bump’.
The results of this study therefore shed light on 1) which motion components of helicopter maneuvers can be properly cued with short-stroke motion systems, and 2) how short-stroke motion systems should be leveraged for helicopter flight simulation. As a recommendation, it can be worthwhile to explore the application of this objective evaluation framework for the tuning of motion cueing algorithms for short-stroke motion systems. In particular, this study revealed false pitch and roll rate cues associated with tilt coordination, which could demand a retuning of the motion cueing algorithm. The objective evaluation framework can then be used as a test method to identify optimal settings. To add, the simultaneous evaluation of the retuned short-stroke motion system with the objective evaluation framework by means of a subjective pilot assessment could further prove the validity of the objective evaluation framework, and could therefore constitute an potential future research direction.
A
Acknowledgement
This research is part of an internally-funded research program at the Royal Netherlands Aerospace Center. We would like to thank the Royal Netherlands Air Force Pilot who flew the MTEs for this research.
Declarations
The authors declare that they have no competing financial or non-financial interests that are directly or indirectly related to the work submitted for publication.
A
Author Contribution
B.J.V. and G.H.J did the literature review together, and G.H.J. wrote the introduction and background. B.J.V. wrote the method, results, discussion, and conclusion, and performed the OMCT and time-domain testing. All authors reviewed the manuscript.
A
Data Availability
The data is not publicly available due to company policy, but can be provided by the authors by request.
7. References
1. Anon.: H125 virtual reality simulator approved by DGAC for type rating. Airbus Helicopters Collaborative Space for Customers and Partners. https://airbusworld.helicopters.airbus.com/web/guest/home#/detail/619109682 (2025). Accessed 5 June 2025
2. Anon.: Los Angeles Police Dept puts VR simulator to work. Aviation International News. https://www.ainonline.com/aviation-news/business-aviation/2025-03-03/los-angeles-police-put-vr-simulator-work (2025). Accessed 5 June 2025
3. Martini, T.: Evaluation of virtual and mixed reality technologies in helicopter simulation. In: Proceedings of the Vertical Flight Society 80th Annual Forum & Technology Display, pp. 1–12, Montreal, Canada (2024)
4. Rauschnabel, P.A., Felix, R., Hinsch, C., Shahab, H., Alt, F.: What is XR? Towards a framework for augmented and virtual reality. Comput. Hum. Behav. 133, 107289 (2022). https://doi.org/10.1016/j.chb.2022.107289
5. Dymora, P., Kowal, B., Mazurek, M., Romana, S.: The effects of virtual reality technology application in the aircraft pilot training process. IOP Conf. Ser.: Mater. Sci. Eng. 1024(1), 012099 (2021). https://doi.org/10.1088/1757-899X/1024/1/012099
6. Marron, T., Dungan, N., Namee, B.M., O’Hagan, A.D.: Virtual reality and pilot training: existing technologies, challenges, and opportunities. J. Aviat./Aerosp. Educ. Res. 33(1), 1–18 (2024). https://doi.org/10.58940/2329-258X.1980
7. Oh, C.G.: Pros and cons of a VR-based flight training simulator: empirical evaluations by student and instructor pilots. Proc. Hum. Factors Ergon. Soc. Annu. Meet. 64, 193–197 (2020). https://doi.org/10.1177/1071181320641047
8. Anon.: CS-FSTD(H): Certification specifications – flight simulation training devices (helicopters), initial issue. European Union Aviation Safety Agency, Cologne (2012)
9. De Winter, J.C.F., Dodou, D., Mulder, M.: Training effectiveness of whole body flight simulator motion: a comprehensive meta-analysis. Int. J. Aviat. Psychol. 22(2), 164–183 (2012). https://doi.org/10.1080/10508414.2012.663247
10. Kim, J., Hwang, J., Park, T.: Effect of motion cues on simulator sickness in a flight simulator. In: Virtual, Augmented and Mixed Reality. Design and Interaction, 12th Int. Conf., pp. 31–41, Copenhagen, Denmark (2020). https://doi.org/10.1007/978-3-030-49695-1_3
11. Calderon, C.M., Bae, J., Barnett-Cowan, M., Cao, S.: Navigating simulator sickness: the effect of flight maneuvers in fixed-base flight simulators. Proc. Hum. Factors Ergon. Soc. Annu. Meet. 68(1), 456–460 (2024)
12. Chang, E., Kim, H., Yoo, B.: Virtual reality sickness: a review of causes and measurements. Int. J. Hum.-Comput. Interact. 36(11), 1027–1051 (2020). https://doi.org/10.1080/10447318.2020.1778351
13. Anon.: FSTD special conditions for the use of head-mounted displays (HMD) combined with a motion platform with reduced motion envelope. European Union Aviation Safety Agency, Cologne (2023)
14. Anon.: FAA Part 60: Flight simulation training device initial and continuing qualification and use. Federal Aviation Administration, Washington, DC (2008)
15. Hosman, R.J.A.W., Advani, S.K.: Design and evaluation of the objective motion cueing test and criterion. Aeronaut. J. 120(1227), 873–891 (2016). https://doi.org/10.1017/aer.2016.35
16. Goldberg, J.M., Fernández, C.: Vestibular mechanisms. Annu. Rev. Physiol. 37(1), 129–162 (1975)
17. Reid, L.D., Nahon, M.A.: Flight simulation motion-base drive algorithms: Part 1 – development and testing of the equations. UTIAS Report No. 296, University of Toronto Institute for Aerospace Studies, Toronto (1985)
18. International Civil Aviation Organization: Doc. 9625 – Manual of Criteria for the Qualification of Flight Simulation Training Devices, vol. II: Helicopters, attachment F, 1st edn. ICAO, Montréal (2012)
19. Royal Aeronautical Society: Aeroplane Flight Simulation Training Device Evaluation Handbook, vol. 1: Objective Testing, 4th edn. RAeS, London (2009)
20. Stroosma, O., Van Paassen, M.M., Mulder, M., Hosman, R.J.A.W., Advani, S.K.: Applying the objective motion cueing test to a classical washout algorithm. In: Proceedings of the AIAA Modeling and Simulation Technologies Conference, pp. 1–14, Boston, USA (2013). https://doi.org/10.2514/6.2013-4834
21. Advani, S.K., Hosman, R.J.A.W.: Towards standardising high-fidelity cost-effective motion cueing in flight simulation. In: Proceedings of the Royal Aeronautical Society Conference: Cutting Costs in Flight Simulation, pp. 45–56, London, UK (2006)
22. Advani, S.K., Hosman, R.J.A.W., Potter, M.: Objective motion fidelity qualification in flight training simulators. In: Proceedings of the AIAA Modeling and Simulation Technologies Conference and Exhibit, pp. 1–12, AIAA, Washington, DC (2007)
23. Baskett, B.: ADS-33E-PRF: Aeronautical design standard, performance specification, handling qualities requirements for military rotorcraft. US Army Aviation and Missile Command, Redstone Arsenal, AL (2000)