Synergistic Visual Calibration (SVC)

• SVC is a sensory-conflict framework that predicts, quantifies, and suppresses motion sickness by aligning visual and vestibular inputs.
• The model integrates multi-channel sensory dynamics with a closed-loop, parameterized visual control law to compute and mitigate MSI.
• Human experiments validate that SVC-based optimization significantly reduces motion sickness incidence and delays symptom onset.

Synergistic Visual Calibration (SVC) refers to a sensory-conflict framework for predicting, quantifying, and suppressing motion sickness, particularly in dynamic transport environments, by optimally generating visual stimuli that reduce the perceptual conflict between vestibular and visual cues. The SVC model, anchored in subjective-vertical conflict theory, models motion sickness as a result of dynamic discrepancies between multi-sensory signals informing the brain’s estimate of upright orientation. Recent research demonstrates the closed-loop application of SVC-based modeling to optimally synthesize visual angular velocity inputs, thereby reducing motion sickness incidence (MSI) in both simulation and human experiments (Tamura et al., 2023).

1. Subjective-Vertical Conflict Theory: Multi-Channel Sensory Dynamics

SVC theory posits that motion sickness stems from the cumulative conflict between the sensed gravito-inertial vertical (driven by vestibular and visual afferents) and the internal neural estimate of vertical. In the explicit model, parallel vestibular channels (otolith organs and semicircular canals) and a visual channel provide inputs. Each sensory channel consists of both a direct measurement path and an internal model that predicts the expected sensory output given self-motion states. The main signal flow comprises:

• Gravito-inertial acceleration, $f(t) = g + a(t)$, where $g$ is the gravitational vector and $a(t)$ is measured by the otoliths.
• Otolith channel transmits $f(t)$ with unity dynamics.
• Semicircular canal (SCC) dynamics in the Laplace domain: $$\omega_s(s) = H_{SCC}(s)\omega(t) = \frac{\tau_a s}{\tau_a s + 1}\omega(t)$$, where $\tau_a \approx 0.5$ s.
• Low-pass neural fusion: $\dot{v}_s(t) = f(t) - v_s(t)$, estimating vertical.
• Visual channel (VIS): provides angular velocity, $\omega_{vis}(t)$, derived from optical flow, with identity dynamics.
• Internal models mirror sensory dynamics; e.g., $\overline{SCC}(s) = SCC(s)$.
• Feedback (error) signals such as $A_\omega$, $g$0, $g$1, and $g$2 (the latter for visual feedback) are computed as linear differences (with gains) between actual and predicted states.

A summary of parameter values is given below:

Parameter	Value	Role
$g$3	0.1	Vestibular angular feedback gain
$g$4	0.8	Vestibular acceleration feedback
$g$5	5.0	Vestibular vertical feedback gain
$g$6	7.0	Visual feedback gain
$g$7	0.5 s	SCC time constant
$g$8	0.5 m/s²	Hill-function half-saturation
$g$9	7%	Hill-function maximum output
$a(t)$0	10 min	MSI accumulation time constant

Conflict signals update the internal state estimates, closing the sensory-prediction feedback loop.

2. Computational Model of Motion Sickness Incidence (MSI)

With the vertical conflict signal $a(t)$1 established, MSI is computed through a nonlinear, temporally-accumulating process:

• Absolute vertical conflict $a(t)$2 is mapped by a static Hill function: $a(t)$3, where $a(t)$4, $a(t)$5 m/s².
• The Hill output is integrated via a second-order lag: $a(t)$6, with $a(t)$7 min, $a(t)$8.
• $a(t)$9 represents the expected percentage of a subject group reaching emesis under the stimulation up to time $f(t)$0.

This modular model allows for direct linkage from head or vehicle kinematics and visual states to population sickness outcomes.

3. Optimization of Visual Inputs to Mitigate MSI

SVC-based suppression of motion sickness requires the generation of visual angular velocities $f(t)$1 that minimize MSI over a known or predicted motion horizon. The functional optimization is non-trivial; a parameterized control law is used in practice:

• Visual input model:

$f(t)$2

with $f(t)$3.

• The cost function is $f(t)$4, with $f(t)$5 as the trial endpoint.
• Optimal parameters $f(t)$6 are found by offline minimization of $f(t)$7 over known sinusoidal vehicle acceleration.
• During deployment, $f(t)$8 is measured, $f(t)$9 computed, and fed to the head-mounted display (HMD) in real time.

This approach operationalizes SVC theory as an explicit method for closed-loop, optimal visual-feedback synthesis in dynamic environments.

4. Human Experimental Protocol and Quantitative Outcomes

Experimental validation involved seven healthy male participants (20s, MSSQ susceptibility 0–73%). Participants experienced fore-aft sinusoidal vehicle motion ($$\omega_s(s) = H_{SCC}(s)\omega(t) = \frac{\tau_a s}{\tau_a s + 1}\omega(t)$$0 m/s², $$\omega_s(s) = H_{SCC}(s)\omega(t) = \frac{\tau_a s}{\tau_a s + 1}\omega(t)$$1 Hz) in an automated mobility platform with head motion constrained via neck brace. Visual stimuli comprised small spheres displayed on an HMD (VIVE PRO Eye), presented under two conditions:

• Fixed: spheres static ($$\omega_s(s) = H_{SCC}(s)\omega(t) = \frac{\tau_a s}{\tau_a s + 1}\omega(t)$$2)
• Optimized (proposed law): spheres rotate in pitch, driven by real-time $$\omega_s(s) = H_{SCC}(s)\omega(t) = \frac{\tau_a s}{\tau_a s + 1}\omega(t)$$3

Participants reported motion sickness every minute using the MISC (Motion Illness Symptom Classification, 0–10) scale, with cessation if MISC ≥ 5.

Results summary:

Measurement	Fixed Condition	Proposed Law Condition
Simulated MSI after 300s	≈ 3.5%	≈ 1.2%
Group-mean MISC at 15 min	≈ 3.0	≈ 1.8
Subjects with reduced/delayed peak MISC	--	6/7

In simulation, the SVC model predicted substantial reduction in MSI, corroborated in human trials where most subjects demonstrated lower and/or delayed symptom onset under the optimized-visual-feedback protocol (Tamura et al., 2023).

5. Parameter Dependencies, Constraints, and Sensitivity Analysis

The efficacy and stability of SVC-based control are contingent on several parameter choices:

• The visual feedback gain $$\omega_s(s) = H_{SCC}(s)\omega(t) = \frac{\tau_a s}{\tau_a s + 1}\omega(t)$$4 must be selected to balance sufficient error correction without inducing excessive perceptual conflict; an optimal value of $$\omega_s(s) = H_{SCC}(s)\omega(t) = \frac{\tau_a s}{\tau_a s + 1}\omega(t)$$5 was adopted.
• Hill-function half-saturation $$\omega_s(s) = H_{SCC}(s)\omega(t) = \frac{\tau_a s}{\tau_a s + 1}\omega(t)$$6 and time constant $$\omega_s(s) = H_{SCC}(s)\omega(t) = \frac{\tau_a s}{\tau_a s + 1}\omega(t)$$7 set the sensitivity and accumulation rate of sickness prediction; reducing $$\omega_s(s) = H_{SCC}(s)\omega(t) = \frac{\tau_a s}{\tau_a s + 1}\omega(t)$$8 or $$\omega_s(s) = H_{SCC}(s)\omega(t) = \frac{\tau_a s}{\tau_a s + 1}\omega(t)$$9 increases model responsivity to small conflicts.
• The optimization performed offline presupposes a known motion profile; application to general, stochastic vehicle trajectories would require real-time or adaptive optimization (e.g., receding-horizon methods).
• The experimental context—restricted to uniaxial, fore-aft stimulus with highly constrained head motion—limits generalizability to naturalistic, six-degree-of-freedom head-vehicle dynamics.
• Only a narrow demographic (young males) was tested; broader participant pools and statistical analyses are required to establish population-level effects.

6. Limitations and Research Directions

Limitations identified in the current formulation of SVC-based visual calibration include:

• Applicability restricted to single-axis motion and suppressed voluntary head movements, which does not capture the complexity of real-world transport.
• Offline optimization is not scalable to environments with dynamic, unpredictable motion unless reformulated as an adaptive or model-predictive controller.
• Small sample size and narrow demographic representation impact the statistical power and generality of conclusions.

Prospective enhancements outlined for future work include:

• Extension of the SVC-MSI framework to six degrees of freedom (6 DoF) encompassing roll, pitch, and yaw dynamics with online head-motion prediction.
• Implementation of model-predictive control schemes for in-situ, real-time optimization of visual feedback in response to live acceleration measurements.
• Personalization of SVC-MSI parameters ($\tau_a \approx 0.5$0 values, $\tau_a \approx 0.5$1, $\tau_a \approx 0.5$2, $\tau_a \approx 0.5$3) to individual users through calibration drives or laboratory vestibular testing.
• Joint optimization of vehicle and visual motion stimuli for group-level sickness suppression in multi-passenger and mixed-traffic contexts.

This body of work constitutes the first demonstration of real-time, closed-loop optimization of visual angular input leveraging the SVC model to suppress motion sickness symptoms during actual vehicle motion, validated both in silico and through controlled human-subject testing (Tamura et al., 2023).