Tactile Feedback Applied to the Neck to Assist Wheelchair Navigation

This work proposes tactile directional cues applied to the neck to assist in power wheelchair navigation. Haptic directional cues, presented via kinesthetic and tactile guidance to the arms and hands, have proven to help users better navigate smart power wheelchairs. However, this type of guidance would not be usable by people who experience both a loss of motor function and a loss of the sense of touch of their upper extremities, whereas tactile directional cues provided through the neck can potentially assist these individuals. To evaluate the neck as a venue for providing directional tactile cues, we created two wearable tactile devices for the neck; a novel skin-stretch device and a vibrotactile collar. 

We conducted a within-subject user study in which 14 participants were asked to navigate a virtual power wheelchair through an indoor environment consisting of hallways and doors. Participants completed the task with guidance cues from the skin-stretch device, guidance cues from the vibrotactile device, and no guidance cues. Measures of the participant’s performance during the user study show that tactile guidance through the neck with both skin-stretch cues and vibrotactile cues can help users better navigate a power wheelchair. Subjective measures of the participant’s responses to questionnaires indicate that vibrotactile cues are preferred over skin-stretch cues. 

More information about the haptic devices:

Skin-Stretch haptic device: The skin-stretch haptic feedback device is comprised of two mirrored parts, one for the left side of the neck and the other for the right side of the neck. Each part is driven by a servo motor (Tower Pro MG90S) that is adhered to a 3D printed mount, which is taped to the back of the neck via skin-safe tape (3M™ Red Dot™ Monitoring Electrode). The tape at the back of the neck is referred to as the ‘fixed tape’ because it remains relatively still when the device is in operation. The servo motor drives a two-bar linkage. The end of the two-bar linkage is adhered to the side of the user’s neck using another piece of skin-safe tape. The tape attached to the end of the two-bar linkage is referred to as the ‘tactor’. The tactor is the part of the device that will travel to stretch the skin. During operation, the tactor begins in a neutral position and does not apply any shear force to the skin. A stretch cue is generated by moving the tactor from the neutral position in either the forward or backward direction. Pilot testing showed that displacing the tactor approximately 12mm from the neutral position resulted in a stretch sensation that was both noticeable and comfortable. 

Vibrotactile haptic device: The vibrotactile haptic feedback device, shown in the bottom row of Fig. 1, is a wearable collar that contains eight ERM motors (BestTong Mini-Vibration Motors 10mmx3mm DC 3V 12000RPM/200 hz). The collar is fit to the user via an adjustable Velcro fastener. 

System Control: The servo motors of the skin-stretch and each ERM motor was controlled individually using a Raspberry Pi model 3B+ and custom circuitry. The motors and the Raspberry Pi were powered via a battery-powered 6V power supply. 

Power wheelchair simulation:

Power wheelchair simulation: In order to test the utility of haptic feedback delivered to the neck in the context of power wheelchair navigation, we created a power wheelchair simulator using the CoppeliaSim software. We imported this model into CoppeliaSim and added realistic physical properties to the wheelchair. We also added controllable rotational joints to the chair’s front wheels. In the wheelchair simulator, the human driver uses a joystick (Logitech Extreme 3D Pro) to command the linear velocity, and the angular velocity of the wheelchair. 

A virtual camera is rigidly attached to the chair to provide the user with a first person view that would be similar to the view of a person seated in the chair. The user can see both the left and right arms of the chair from this view. The simulator updates at a rate of 20hz and haptics commands were  generated at a rate of 8hz.