LUMIN: an open source dynamic reconfigurable lightboard for neurorehabilitation and athletic training

Graphical abstract

Abstract

Lightboards are commonly used in neurorehabilitation and athletics to assess improved volitional task function, hand-eye coordination, spatial awareness, and reaction time. Current commercial products are expensive, large, static, and not open sourced. In this work, we present a Lightboard Unit for Movement and Integrated Neurorehabilitation (LUMIN), a low cost, open source, and accessible lightboard for neurorehabilitation and athletic training alike. LUMIN uses a motorized mechanical superstructure that allows for rotational movement and radial expansion of the interactive LED components. These capabilities allow the device to adapt to each individual user’s anthropometry and tasks. The device uses an Arduino based microcontroller, stepper motor controlled linear and rotational mechanisms, and tactile Push Button RGB LEDs to allow for adjustable interactive programming. The device was fabricated using commercially sourced components and materials, costing approximately $900. Performance of the device demonstrates controlled linear and angular movements, low LED latency, and a workspace accessible for a broad range of user wingspans while remaining lightweight and portable compared to existing commercial systems. All of the design files were released under an open-source license to promote reproducibility and further research in neurorehabilitation and eventually human performance training. The goals for this device were to construct and test system performance. While future studies will use the lightboard to compare performance from a user-perspective vs. existing systems.

1. Hardware in context

We designed a Lightboard Unit for Movement and Integrated Neurorehabilitation (LUMIN) to train volitional task function, hand-eye coordination, spatial awareness, and reaction time. This technology is typically used as a static system in post-stroke rehabilitation or sports training environments [1], [2]. The presented technology will be used for research to create a more dynamic training environment and to enable additional modalities not available to traditional static light boards. (See Table 1).

Table 1.

Specifications table.

Hardware name	Lightboard unit for movement and integrated neurorehabilitation (LUMIN)
Subject area	Rehabilitation Devices
Hardware type	Neurorehabilitation
Closest commercial analog	DynaVision D2, FITLIGHT Vision Board, Vision Coach
Open-source license	Creative Commons Attribution-NonCommercial-NoDerivates (CC BY-NC-ND)
Cost of hardware	$900.00
Source file repository	https://doi.org/10.5281/zenodo.18940336

Current lightboards are large (4′ x4′), heavy (300 + lbs. in some cases), and expensive (>$10,000 per unit), thus making them inaccessible for many potential users. Some of the existing systems include the DynaVision D2 Light Board, FITLIGHT Vision Board, and Vision Coach. The DynaVision D2 is a 48″ x 48″ board that is on a mobile stand allowing it to be mobile within a room [3], [4]. The Vision Coach is a 50″ x 34″ interactive board that is mounted to a wall where the board can slide up and down adjusting to the patient's height [5]. Finally, the FITLIGHT Vision Board is a series of puck lights that can be placed in various positions on the board where the patient can interact with them [6], [7]. Augmented and Virtual Reality systems are currently being researched but have not yet been developed into commercial products [8], [9].

The presented LUMIN board is broadly more affordable than the light boards depicted in Table 2. One exception is the FITLIGHT system which is only ∼$500 cheaper than the LUMIN [10]. However, the listed price for the FITLIGHT is for their base system which only includes four puck lights. To use FITLIGHT with a comparable configuration to the LUMIN, additional puck lights, magnetic vision board, and colored covers would need to be added, with a total cost of ∼$2,400. Conversely, the entire LUMIN device costs $900 as depicted. This is also significantly more cost effective than the >$10,000 price tag of the DynaVision D2 and Vision Coach systems. LUMIN also provides many features that the other devices do not. For example, LUMIN’s hardware layout and LED count is modular and customizable, it is completely open source to allows for programmable task or light patterns, dynamic movement training and workspace expandability, and fully accessible data gathering and export. However, the LUMIN board does not offer Specialized Visual Lenses training like the Vision Coach.

Table 2.

LUMIN vs. Current commercial products.

Feature category	LUMIN	DynaVision D2	FITLIGHT Vision Board	Vision Coach
Approx. Cost	∼$900 (build cost)	>$10,000	∼$1,399	>$10,000
Hardware Layout	Modular, reconfigurable	Fixed board	Fixed pods	Fixed panel
LED Count / Nodes	Customizable	Fixed	Fixed	Fixed
Dynamic Movement	Yes	No	No	No
Reaction Time Measurement	Yes	Yes	Yes	Yes
Multi-modal Interaction	Yes (visual + motor)	Yes	Yes	Yes
Specialized Visual Lenses	No	No	No	Yes
Software Customization	Full (open-source)	None	Limited	Limited
Expandability	Yes	No	No	No
Repairability	Off-the-shelf parts	Vendor-only	Vendor-only	Vendor-only
Data Export	Custom	Limited	Limited	Limited
Intended Use	Research, rehab, training	Clinical, training	Athletic training	Vision training

Previous studies have shown that boards like Vision Coach have led to positive improvements in basic tasks of day to day activities, such as driving [11], [12]. Lightboards also have valuable use with athletes as well as physical and occupational therapy patients [13], [14]. It is used to test and train reaction time with proven positive correlation between Light Board training and reaction time improvement [15], [16]. This means that if LUMIN can provide a more accessible Light Board, patients and athletes alike can benefit from broader access to a technology that was previously restricted to specialized environments [17].

2. Hardware description

The proposed design consists of several sub-systems including a mechanical superstructure, motorized configuration manager, electronic controller, and user interface. The device has several LED buttons that act as a user interface (i.e., users must click the button when it illuminates), and the system can reposition the LED buttons through a series of rotations and linear translations. The device can extend the LED positions to allow buttons to be positioned over a large area of the user’s workspace (wingspan).

Mechanical Superstructure. Fig. 1 shows the mechanical superstructure from various perspectives and breaks down the main mechanical systems within the device’s structure. The structure has four primary areas. The aluminum extrusion frame, the Rotational Mechanism system, the Bevel Gear Box, and the Lead Screw Assembly System. The aluminum extrusion frame is constructed out of 10 series T Slot Aluminum Extrusion to ensure a strong yet lightweight frame. The frame has one ¼“ steel center axle that is located at the top of the frame that the device rotates around.

Fig. 1.

Mechanical systems overview.

Motorized Configuration Manager. The Rotational Mechanism system provides the LUMIN with its ability to rotate and to control coordinate localization within a workspace grid for the light board. The mechanism uses NEMA 23 Stepper Motors to rotate spur gears around a stationary spur gear attached to the aluminum extrusion frame. The housing for NEMA 23 is in the center matrix housing. This structure also houses the Arduino MEGA 2560, four DM320T stepper drivers, two DM556T stepper drivers, four NEMA 17 Stepper Motors, and four bevel gear boxes. The center matrix structure has a pinwheel shaped structure that holds together the frame of the center matrix. On each side of the pinwheel, there are two built-in mounting points for each of the NEMA 23 allowing the structure to rotate.

LUMIN has four modular control arms that each host a “Pie Slice” which contains the Push Button RGB LEDs comprising the user interface. Each slice is extended and contracted using a NEMA 17 Stepper Motor that is attached to the bevel gear box. This bevel gear box was designed to save space in the center matrix structure.

The final portion of the mechanical system of the device is the lead screw assembly system. This is the system that allows the pie slice on the control arms to expand and contract, as well as hold the pie slice the patient interacts with. The system connects to the bevel gear box which rotates the lead screw. On the lead screw there is a top and bottom stationary piece that connect to the nut which prevent the nut from rotating freely on the bottom and holds the pie slice in place on the top.

Electronic Controller. The electronic controller is comprised of three main components: a microcontroller, stepper drivers, and stepper motors. LUMIN uses an Arduino MEGA 2560 microcontroller to collect data from the user interface and to regulate the components. The Arduino controls four DM320T 2-Phase digital stepper drivers, two DM556T 2-phase digital stepper drivers, four NEMA 17 stepper motors, two NEMA 23 stepper motors, ten 44 mm Push Button LEDs, and ten 60 mm Push Button LEDs. To power these components requires two different levels of power, a 5 V source to power the Arduino and Push Button LEDs, and a 24 V source to power Stepper Motors and their Drivers (Fig. 2).

Fig. 2.

Electrical system overview.

Microcontroller. The Arduino MEGA 2560 was chosen as the microcontroller due to the low cost and 54 digital I/O pins that could be used to connect to all the required components. In total, we required 52 digital I/O.

Stepper Motors. The device has four NEMA 17 Bipolar 0.9° 1.6 A stepper motors and two NEMA 23 1.8° 2.8 A stepper motors. To determine the motor requirements, we design motion characteristics of the system, and estimate the desired motor torque,

Total Inertia of Rotating Portion of Device.

The total system inertia will define the motion characteristics and the torque needed to produce the motion. The system’s inertia was estimated by breaking the device down into four rectangles and one square, to represent the arms and center respectively. The total moment of inertia ($I$) was calculated by eq. (1). Where $M$(lbs.) is the total mass of the portion of the device and $r$ is the distance from the center to the furthest point of each portion of the device. The total inertia ($I$) is then shown in eq. (2).

$$MOI\left(I\right)=\mathrm{\Sigma }Mass\left(M\right)·RadiusfromCenter(r^2)=4(M_{\mathit{arm}}{r}_{\mathit{arm}}^2)+M_{\mathit{center}}{r}_{\mathit{center}}^2$$ (1)

$$I=\mathrm{\Sigma }M{r}^2=4\left(\left(3lbs\right)\left({\left(22+\frac{11}{2}\right)}^{2}in\right)\right)+\left(7lbs\right){\left(\frac{11}{2}in\right)}^2=9286.75lbi{n}^2=2.73kg{m}^2$$ (2)

Angular Velocity and Acceleration

We designed the system so that it can rotate 180° ($\pi$ radians) in approximately 3 s, this was chosen as it is an adequate speed to test users over multiple programs. Thus, eqs. (3), (4) define the required angular velocity and acceleration of the system, respectively.

$$AngularVelocity\left(\omega \right)=\frac{\mathit{ChangeinAngle}(\mathrm{\Delta }\theta )}{\mathit{Changeintime}(\mathrm{\Delta }t)}=\frac{\pi radians}{3sec}=\frac{\pi }{3}\frac{\mathit{rad}}{\mathit{sec}}$$ (3)

$$AngularAcceleration\left(\alpha \right)=\frac{ChangeinAngularVelocity\left(\mathrm{\Delta }\omega \right)}{Changeintime\left(\mathrm{\Delta }t\right)}=\frac{{\omega }_f-{\omega }_i}{\mathrm{\Delta }t}=\frac{\frac{\pi }{3}-0}{3}=0.349\frac{\mathit{rad}}{\mathit{se}{c}^2}$$ (4)

Torque.

Using the desired rotational dynamics and the inertia, we then compute the required rotational torque needed at the rotational point, shown in eq. (5).

$$\tau =I\alpha =2.73kg{m}^2(0.349\frac{\mathit{rad}}{\mathit{se}{c}^2})=0.95Nm$$ (5)

Thus, we approximate that a ∼ 1N*m of torque motor would suffice for our system. The NEMA 23 motor was then selected as it had the same wiring set up as the NEMA 17 motor and produced 1.2 N*m of torque. Stronger motors can be substituted to increase the rotational speeds of the system; however, additional mechanical safety may need to be considered in this case.

Stepper Driver. A DM320T Stepper Driver was used due to a current limit of 2.2 A, since the motor required 1.68 A. Connector blocks were used to screw in the wires from the motors and MCU.

The DM556T Stepper driver was used for the NEMA 23 motors, as it can handle up to 5.6 A while the NEMA 23 outputs 2.8 A.

Push Button RGB LEDs. Ten 44 mm and 60 mm square LED Illuminated Push Button LEDs were used. The LEDs provide tactile feedback to the user and are diffused RGB-type LEDs allowing programmable coloration for various types of tasks.

User Interface. The patient user interacts LUMIN by positioning themselves with their eyes leveled with the center of the device. To keep patient distractions to a minimum, the front facade of the device is black, and the LEDs are red when lit to provide maximum contrast.

Advantages.

• •Easily Accessible
• •Modular
• •Affordable

3. Design files summary

Table 3

Table 3.

Design file summary.

Design file name	File type	Open-source license	Location of the file
Al Ex Stopper Tight	CAD (.ipt)	(CC BY-NC-ND)	https://zenodo.org/api/records/18940336/draft/files/Al%20Ex%20Stopper%20Tight.ipt/content
Back Support Structure	CAD(.iam)	(CC BY-NC-ND)	https://zenodo.org/api/records/18940336/draft/files/Back%20Support%20Structure.iam/content
Bearing Holder Plate	CAD(.ipt)	(CC BY-NC-ND)	https://zenodo.org/api/records/18940336/draft/files/Bearing%20Holder%20Plate.ipt/content
Bevel Gear Box Support	CAD(.ipt)	(CC BY-NC-ND)	https://zenodo.org/api/records/18940336/draft/files/Bevel%20Gear%20Box%20support.ipt/content
Center Pie Frame Bottom Left	CAD(.ipt)	(CC BY-NC-ND)	https://zenodo.org/api/records/18940336/draft/files/Center%20Pie%20Frame%20Bottom%20Left.ipt/content
Center Pie Frame Bottom Right	CAD(.ipt)	(CC BY-NC-ND)	https://zenodo.org/api/records/18940336/draft/files/Center%20Pie%20Frame%20Bottom%20Right.ipt/content
Center Pie Frame Top Left	CAD(.ipt)	(CC BY-NC-ND)	https://zenodo.org/api/records/18940336/draft/files/Center%20Pie%20Frame%20Top%20Left.ipt/content
Center Pie Frame Top Right	CAD(.ipt)	(CC BY-NC-ND)	https://zenodo.org/api/records/18940336/draft/files/Center%20Pie%20Frame%20Top%20Right.ipt/content
Extrusion Connector	CAD(.ipt)	(CC BY-NC-ND)	https://zenodo.org/api/records/18940336/draft/files/Extrusion%20Connector.ipt/content
Lead Screw Arm Brace	CAD(.ipt)	(CC BY-NC-ND)	https://zenodo.org/api/records/18940336/draft/files/Lead%20Screw%20Arm%20Brace.ipt/content
Lead Screw Arm Extension	CAD(.ipt)	(CC BY-NC-ND)	https://zenodo.org/api/records/18940336/draft/files/Lead%20Screw%20Arm%20Extension.ipt/content
Lead Screw Coupler Connector Female	CAD(.ipt)	(CC BY-NC-ND)	https://zenodo.org/api/records/18940336/draft/files/Lead%20Screw%20Coupler%20Connector%20Female.ipt/content
Lead Screw Coupler Connector Male	CAD(.ipt)	(CC BY-NC-ND)	https://zenodo.org/api/records/18940336/draft/files/Lead%20Screw%20Coupler%20Connector%20Male.ipt/content
Lead Screw End Support	CAD(.ipt)	(CC BY-NC-ND)	https://zenodo.org/api/records/18940336/draft/files/Lead%20Screw%20End%20Support.ipt/content
Lead Screw System Assembly	CAD(.iam)	(CC BY-NC-ND)	https://zenodo.org/api/records/18940336/draft/files/Lead%20Screw%20System%20Assembly.iam/content
NEMA 17 Gear Plate	CAD(.ipt)	(CC BY-NC-ND)	https://zenodo.org/api/records/18940336/draft/files/Nema%2017%20Gear%20Plate.ipt/content
NEMA 23 Larger Gear Back Plate	CAD(.ipt)	(CC BY-NC-ND)	https://zenodo.org/api/records/18940336/draft/files/Nema%2023%20Larger%20Gear%20Back%20Plate.ipt/content
NEMA 23 Larger Gear	CAD(.ipt)	(CC BY-NC-ND)	https://zenodo.org/api/records/18940336/draft/files/Nema%2023%20Larger%20Gear.ipt/content
NEMA 23 Smaller 1.125 Gear Cross Female	CAD(.ipt)	(CC BY-NC-ND)	https://zenodo.org/api/records/18940336/draft/files/Nema%2023%20Smaller%201.125%20Gear%20Cross%20Female.ipt/content
NEMA 23 Smaller 1.125 Gear Cross Male 20 mm	CAD(.ipt)	(CC BY-NC-ND)	https://zenodo.org/api/records/18940336/draft/files/Nema%2023%20Smaller%201.125%20Gear%20Cross%20Male%2020%20 mm.stl/content
NEMA 23 Smaller 1.125 Gear Cross Male	CAD(.ipt)	(CC BY-NC-ND)	https://zenodo.org/api/records/18940336/draft/files/Nema%2023%20Smaller%201.125%20Gear%20Cross%20Male.ipt/content
Pie Slice Frame Dovetail Female	CAD(.ipt)	(CC BY-NC-ND)	https://zenodo.org/api/records/18940336/draft/files/Pie%20Slice%20Frame%20Dovetail%20Female.ipt/content
Pie Slice Frame Dovetail Male	CAD(.ipt)	(CC BY-NC-ND)	https://zenodo.org/api/records/18940336/draft/files/Pie%20Slice%20Frame%20Dovetail%20Male.ipt/content
Structure Gear Support	CAD(.ipt)	(CC BY-NC-ND)	https://zenodo.org/api/records/18940336/draft/files/Structure%20Gear%20Support.ipt/content
Updated Lead Screw Stationary Piece Extended	CAD(.ipt)	(CC BY-NC-ND)	https://zenodo.org/api/records/18940336/draft/files/Updated%20Lead%20Screw%20Stationary%20Piece%20Extended.ipt/content
Updated Lead Screw Stationary Piece Top	CAD(.ipt)	(CC BY-NC-ND)	https://zenodo.org/api/records/18940336/draft/files/Updated%20Lead%20Screw%20Stationary%20Piece%20Top.ipt/content
Updated Smaller 15 tooth Bevel Gear	CAD(.ipt)	(CC BY-NC-ND)	https://zenodo.org/api/records/18940336/draft/files/Updated%20Smaller%2015%20tooth%20Bevel%20Gear%20Female.ipt/content
Updated Smaller 20 tooth Bevel Gear Female M5 Screw	CAD(.ipt)	(CC BY-NC-ND)	https://zenodo.org/api/records/18940336/draft/files/Updated%20Smaller%2020%20tooth%20Bevel%20Gear%20Female%20 M5%20screw.stl/content
Updated Smaller 20 tooth Bevel Gear Female	CAD(.ipt)	(CC BY-NC-ND)	https://zenodo.org/api/records/18940336/draft/files/Updated%20Smaller%2020%20tooth%20Bevel%20Gear%20Female.ipt/content
Updated Smaller Bevel Gear Housing Top	CAD(.ipt)	(CC BY-NC-ND)	https://zenodo.org/api/records/18940336/draft/files/Updated%20Smaller%20Bevel%20Gear%20Housing%20Top.ipt/content
Updated Smaller Bevel Gear Housing	CAD(.ipt)	(CC BY-NC-ND)	https://zenodo.org/api/records/18940336/draft/files/Updated%20Smaller%20Bevel%20Gear%20Housing.ipt/content
Updated Smaller Bevel Gear Wall Slide	CAD(.ipt)	(CC BY-NC-ND)	https://zenodo.org/api/records/18940336/draft/files/Updated%20Smaller%20Bevel%20Gear%20Wall%20Slide.ipt/content
Updated Smaller Gear Box Assembly	CAD(.iam)	(CC BY-NC-ND)	https://zenodo.org/api/records/18940336/draft/files/Updated%20Smaller%20Gear%20Box%20Assembly.iam/content
Windmill Center Support Structure Extensions	CAD(.ipt)	(CC BY-NC-ND)	https://zenodo.org/api/records/18940336/draft/files/Windmill%20Center%20Support%20Structure%20Extensions.ipt/content
Windmill Center Support Structure	CAD(.ipt)	(CC BY-NC-ND)	https://zenodo.org/api/records/18940336/draft/files/Windmill%20Center%20Support%20Structure.ipt/content
Wire Holder	CAD(.ipt)	(CC BY-NC-ND)	https://zenodo.org/api/records/18940336/draft/files/Wire%20Holder.ipt/content
Base Structure Assembly	CAD(.iam)	(CC BY-NC-ND)	https://zenodo.org/api/records/18940336/draft/files/Base%20Structure%20Assembly.iam/content
Diffused_10_LEDs_Random_Test	Code(.ino)	(CC BY-NC-ND)	https://zenodo.org/api/records/18940336/draft/files/Diffused_10_LEDs_Random_Test.ino/content
Nema_23_Structure_Rotate.ino	Code(.ino)	(CC BY-NC-ND)	https://zenodo.org/api/records/18940336/draft/files/Nema_23_Structure_Rotate.ino/content
LUMIN Lightboard Demonstration Video	Video(.mp4)	(CC BY-NC-ND)	https://zenodo.org/api/records/18940336/draft/files/LUMIN%20LightBoard%20Demostration%20Video.mp4/content
Random_LED_Game_Random_Movement_Strucutre_Rotate_Test	Code(.ino)	(CC BY-NC-ND)	https://zenodo.org/api/records/18940336/draft/files/Random_LED_Game_Random_Movement_Strucutre_Rotate_Test.ino/content
20_Push_Button_LED_Test	Code(.ino)	(CC BY-NC-ND)	https://zenodo.org/api/records/18940336/draft/files/20_Push_Button_LED_Test.ino/content
3_Push_Button_LED_Test	Code(.ino)	(CC BY-NC-ND)	https://zenodo.org/api/records/18940336/draft/files/3_Push_Button_LED_Test.ino/content
KiCAD Light Board Wiring Diagram PNG	Image(.png)	(CC BY-NC-ND)	https://zenodo.org/api/records/18940336/draft/files/KiCAD%20Light%20Board%20Wiring%20Diagram%20PNG.png/content
Random_LED_Game_Strucutre_Rotate_Test	Code(.ino)	(CC BY-NC-ND)	https://zenodo.org/api/records/18940336/draft/files/Random_LED_Game_Strucutre_Rotate_Test.ino/content
Structure_Rotate_Arms_Rotate_Test	Code(.ino)	(CC BY-NC-ND)	https://zenodo.org/api/records/18940336/draft/files/Structure_Rotate_Arms_Rotate_Test.ino/content
Push_Button_LED_Test	Code(.ino)	(CC BY-NC-ND)	https://zenodo.org/api/records/18940336/draft/files/Push_Button_LED_Test.ino/content
DM320T_Driver_Test	Code(.ino)	(CC BY-NC-ND)	https://zenodo.org/api/records/18940336/draft/files/DM320T_Driver_Test.ino/content
Automatic_Random_LED_Game_Strucutre_Rotate_Test	Code(.ino)	(CC BY-NC-ND)	https://zenodo.org/api/records/18940336/draft/files/Automatic_Random_LED_Game_Strucutre_Rotate_Test.ino/content
Manual_Motor_Control	Code(.ino)	(CC BY-NC-ND)	https://zenodo.org/api/records/18940336/draft/files/Manual_Motor_Control.ino/content
Diffused_LED_Test	Code(.ino)	(CC BY-NC-ND)	https://zenodo.org/api/records/18940336/draft/files/Diffused_LED_Test.ino/content
LightBoard_Simon_Game_InProgress	Code(.ino)	(CC BY-NC-ND)	https://zenodo.org/api/records/18940336/draft/files/LightBoard_Simon_Game_InProgress.ino/content
Case_Study_Expirement	Code(.ino)	(CC BY-NC-ND)	https://zenodo.org/api/records/18940336/draft/files/Case_Study_Expirement.ino/content

4. Bill of materials summary

Table 4

Table 4.

Bill of materials summary.

Designator	Component	Number	Cost per unit −currency (USD)	Total cost −Currency (USD)	Source of materials	Material type
Aluminum Extrusion Corner Brace	30, 10 Series Extrusion Corner Bracket	1	$26.88	$26.88	https://www.amazon.com/1010-Aluminum-Extrusion-Connector-Accessories	Metal
¼” Shaft Collar	8, ¼” Shaft Collar	1	$10.89	$10.89	https://www.amazon.com/dp/B0B4JQ1RPG	Metal
5.05 mm Shaft Collars	10, 5.05 mm Shaft Collars	1	6.99	6.99	https://www.amazon.com/uxcell-Collars-5-05mm-Landing-Stopper	Metal
5 mm to 5 mm Coupler	12, 5 mm to 5 mm Coupler	1	13.99	13.99	https://www.amazon.com/dp/B0CJDVZPC1	Metal
DM556T Stepper Driver	Stepper Driver 1.8–5.6 A, Nema 23	2	30.19	60.38	https://www.omc-stepperonline.com/digital-stepper-driver	Electronic
NEMA 23 Stepper Motor	Stepper Motor 1.24 Nm 2.8A NEMA 23	2	17.82	35.64	https://www.omc-stepperonline.com/nema-23	Motor
NEMA 17 Stepper Motor	Stepper Motor 0.44 Nm 1.68 A NEMA 17	4	14.98	59.82	https://www.omc-stepperonline.com/nema-17	Motor
6.35 mm to 8 mm Shaft Coupler	6.35 mm to 8 mm Shaft Coupler	1	8.99	8.99	https://www.amazon.com/Befenybay-Coupling-Diameter-Aluminum-6-35x8mm	Metal
DM320T Stepper Driver	Stepper Driver 0.3–2.2 A, NEMA 17	4	16.61	66.44	https://www.omc-stepperonline.com/digital-stepper-driver	Electronic
ABS Sheeting	48″ x 48″ 1/16″ Black ABS	2	43.30	86.60	https://www.grainger.com/product/Plastic-Sheet-0-0625-in	Plastic
60 mm Push Button RGB LED	LED Illuminated Push Button 60 mm	10	4.05	40.50	https://www.adafruit.com/product/6261	Electronic
44 mm Push Button RGB LED	LED Illuminated Push Button 44 mm	10	3.15	31.50	https://www.adafruit.com/product/6274	Electronic
RGB LED	Diffused RGB LED − Common Anode	20	1.75	35.00	https://www.adafruit.com/product/159	Electronic
12 ft, 10 Series Aluminum Extrusion	Aluminum Extrusion	2	37.39	74.78	https://www.grainger.com/product/80–20-Framing-Extrusion-10-Series-	Metal
4, Elegoo PLA Filament 1 KG	PLA Filament	1	42.46	42.46	https://www.amazon.com/ELEGOO-Filament	Plastic
End Connector Screws	End Screw Connectors	8	1.70	13.60	https://www.grainger.com/product/80–20-Framing-Extrusion-10-Series-	Metal
Assorted M3 Screws	M3 Screw	2	7.99	15.98	https://www.amazon.com/Metric-Screws	Metal
360 ft, 22 AWG Wire in 6 colors, Flexible Silicone Tinned Copper Electrical Cable	Flexible Wires	1	24.99	24.99	https://www.amazon.com/dp/B0CMLGS4YG	Metal
360 ft, 22 AWG Solid Core Wire in 6 Colors	Solid Core Wires	1	15.29	15.29	https://www.amazon.com/dp/B07TX6BX47	Metal
24 in x 48 in Clear Acrylic Sheet	Acrylic Sheet	1	35.48	35.48	https://www.homedepot.com/p/OPTIX-24-in-x-48-in-x-0–118-1–8-in-Clear-Acrylic-Sheet	Plastic
2, 550 mm Tr8x8 Lead Screws with T8 Brass Nut	Lead Screws	2	24.99	49.98	https://www.amazon.com/350 mm-Tr8X8-Printer-Machine-Thread	Metal
Arduino MEGA 2560	Motor controller	1	49.90	49.90	https://www.amazon.com/Arduino-ATmega2560	Electronic
680 Piece Assorted M5 Nuts and Bolts	M5 Screws	1	18.77	18.77	https://www.amazon.com/dp/B0CJ8XHH78	Metal
NinjaTek Cheetah TPU Filament	TPU Filament	1	89.45	89.45	https://www.amazon.com/NinjaTek-3DCH01129010-Cheetah-Filament-	Plastic

5. Build instructions

Mechanical Superstructure.

The frame is constructed out of 10 series T Slot Aluminum Extrusion. The frame is held together by taping each end of the extrusion and inserting an end screw connector that is the slid into the slot within the extrusion. As the extrusion was cut by hand to ensure a 90-degree angle at every joint, a corner brace was used as shown above in Fig. 3. The ¼“ steel center axle that is located at the top of the frame. This was created by drilling a ¼” hole into the end of the extrusion and then friction-fitting the ¼“ axle into place. The device then uses this axle as its’ point of rotation. One-inch Aluminum extrusion extensions were 3D printed from PLA and attached to the ends of the center matrix housing, as shown in Fig. 3. At the end of each aluminum extrusion arm where the Lead Screw End Support is there is an Al Ex Stopper that is printed out of TPU which is then placed and twisted into the extrusion to keep the end support in place. The Stopper is also used to keep the Bevel Gear Box Support in place on the aluminum extrusion within the Center Matrix Housing. The TPU Stopper is not in any figures but is included within the design files. With the presented configuration, the frame can have some sway when a user interacts with the Push buttons (depending on the force). To counteract this, the frame can be mounted to a wall, or the structure can be reinforced. For example, the base frame can be reinforced with two or more supports that connect the top of the center Aluminum Extrusion in a triangular or similar configuration.

Fig. 3.

Device: real images.

The Rotational Mechanism includes the center rotation gear and the TPU spur gear. The TPU spur gear is comprised of a two-piece (male and female) construction. The male portion of the gear is a shaft that has a cross key which fits into the female portion of the gear, as shown in Fig. 3. The male portion is 3D printed using PLA in a horizontal orientation to ensure that the print layers allowed for greater strength. A vertical printed orientation weakens the geometric layout and can cause structural failures. The female portion of the gear is printed out of 95A TPU. Printing the female portion in PLA can cause the gear to slip and crack at points due to the vector of load exertion. There are two of these gears assembled on opposite points of the center rotation spur gear. The center rotation gear was printed horizontally out of PLA in two pieces for ease of assembly and quality of print. This was performed to minimize the need for support materials that could have compromised the geometry of the gear teeth. The gear was then press fitted onto the ¼“ axle so that it was against the piece of aluminum extrusion. The smaller spur gear was attached via coupler to the NEMA 23, so the spur gear and stationery gear lined up to rotate smoothly. A center rotation gear support was printed, attached to the aluminum extrusion, and inserted to the center rotation gear to prevent it from rotating around the ¼” axle, as shown in Fig. 3.

The housing for NEMA 23 is in the center matrix housing. The center matrix housing is a 13.5″ x 13.5″ square made of aluminum extrusion with a “pinwheel shaped” structure in the center as shown previously in Fig. 3. The pinwheel structure is attached to the center matrix housing by an end connector screw that was taped into the end of the center structures’(pinwheels) arm which slide into the extrusion. Then each piece of extrusion was connected to the corners with a 90-degree corner bracket to ensure the center matrix structure was square. As circled and labeled in Fig. 3, there is a circular plate that is placed in the center of the pinwheel. This is the bearing plate that is press fitted with a ¼“ needle roller bearing to help reduce friction when the device is rotating. The plate is bolted to the pinwheel with four M5-type bolts that go through the frame and are secured on the back with an M5 nut. On each side of the pinwheel, there are two built-in mounting points for each of the NEMA 23 allowing the structure to rotate.

As mentioned above, the gear box is composed of five pieces: the housing, wall slide, top piece, 15 tooth bevel gear, and 20 tooth bevel gear. Every component can be 3D printed out of PLA or ABS except for the bevel gear axles. The first variation of the bevel gear box focuses on getting the correct fit between the housing, wall slide, and top piece. The tolerance for an appropriate fit should be 0.017″ between the housing and wall slide – this allows the fit to be tight enough but without deforming the housing. The wall slides fit into two tracks on the side walls of the housing, and the top piece has two clips that lock into two notches towards the top of the gear box housing. Most design challenges, however, arose from the gears. The gear head was printed out of PLA in a vertical layout position and an M5 screw was screwed into the head, while the side of the screw was filed down to create a flat side that allowed the coupler to make solid connection with the gears. Due to the upward forces and the wall being able to slide up, four M3 screws were added to the corners of each gear box to ensure the gears stayed in place to allow for optimal force transfer when in use. The bevel gear box is screwed into the bevel gear box support, not displayed but available in the design files, which slides into the grooves of the inside of the aluminum extrusion in the Center Matrix Housing.

The lead screw system includes a coupler that connects to the bevel gear box along with a sleeve that attaches to the lead screw. The lead screw is 550 mm with a T8 ACME threading. The sleeve that attaches to it is printed out of TPU and uses friction to hold it in place. It also has a small connector that locks into the bottom of the sleeve and attaches to the coupler on the other end. The lead screw includes a nut with four predrilled holes that were used to hold the system together. The base of the lead screw system is press fitted around the collar of the lead screw nut and has two rails that run down the side while riding along the aluminum extrusion arm to prevent the system from rotating. This configuration also allows the nut to move up and down the lead screw. At the end of each support arm there is a dovetail joint that slides into a joint socket to provide support for the sides of the pie slice. The top piece of system is also press fitted to the lead screw nut and #6 screws are screwed into each of the four predrilled holes of the nut to hold the top and bottom assembly together. From the top piece, the pie slice is secured using M3 screws. The support arms are held together to the pie slice using a strip of Velcro on each side to prevent the slice from sliding along the braces and to allow for easy future maintenance.

The final portion of the mechanical system is the pie slice (user interface). The pie slice has a bottom and top layer made from 1/16″ thick black ABS sheeting with a 3D printed 55 mm tall scaffolding, shown in Fig. 4, Fig. 5, Fig. 6. That allows just enough room for the Push Button LED mechanism to be seated. The scaffolding is shaped to match the pie slice except for the far ends and covers the entire shape of the center square, this can be seen in the design files. The scaffolding was printed out of PLA in two separate pieces for the pie slices and four separate pieces for the center, due to not having a large enough printer to make one piece. There are multiple dovetail joints along the intersection of the scaffolding pieces, as highlighted in Fig. 6 The top layer of ABS has 4 1-¼“ holes cut out for two 44 mm and two 60 mm Push button LEDs to be inserted and screwed in using the hardware that comes with the LEDS, also shown in Fig. 5. The top and bottom ABS layers are held to the scaffolding by inserting M3 screws through the ABS into the pilot hole printed into the scaffolding.

Fig. 4.

Left to right: bevel gear box, internal pie slice view, final device front view, final device isometric view.

Fig. 5.

Lead screw system overview.

Fig. 6.

Finished device images: top left: device front view, top middle: front wiring view, top right: end pie slice scaffolding, bottom left: device isometric view, bottom middle: center piece scaffolding, bottom right: side pie slice scaffolding.

Due to the inclusion of moving mechanical components, appropriate safety precautions should be considered during assembly and operation of the device. It is important that all structural and mechanical components should be securely fastened to ensure safety and stability. When operating the device avoid contacting the rotating arms, gears, and lead screw assemblies, and ensure the device is placed on a stable flat surface and/or securely mounted. Proper maintenance should be kept up on mechanical systems to prevent failures.

Electrical system.

LUMIN’s electrical system is composed of four sections. The microcontroller, stepper motor drivers, stepper motors, and push button RGB LEDs. Fig. 7 shows the layout of how the device was wired. In Fig. 7, the RGB LED is only wired to the red cathode, but this can be reconfigured to enable other colors to be presented. For example, the cathode pins to the blue and green lights on the LEDs are not connected in Fig. 7, but by connecting these the lights can be configured to different colors. The two NEMA 23 and four NEMA 17 stepper motors are routed to their specific drivers. The driver then connects to the microcontroller for the pulse and direction input. The drivers also receive a 24 V source of power from a separate DC power supply. The power supply can only power two of the motors at a time safely, so three separate 24 V sources are used. The microcontroller and stepper motor drivers are housed in the center matrix structure and are attached to a piece of 1/8″ thick acrylic using M3 screws and the original mounting points that came on the components. This can be seen displayed in Fig. 3. The microcontroller provides 5 V power to most components, while it receives power from a USB cable plugged into a computer. The LEDs were placed evenly in the four pie slices and the center structure. Each of the wires from the separate buttons were braided together to keep the wiring organized and prevent them from getting caught in any moving parts.

Fig. 7.

Electrical wiring diagram.

The device utilizes multiple external power supplies and electrical wiring, it is important that appropriate precautions should be taken to reduce possible electrical hazards. Power sources should remain off and disconnected during assembly, wiring, or modification to prevent electrical shock or unintended actuation. Electrical connections should be properly insulated and organized to minimize short circuits and interference with the mechanical systems. An accessible method of disconnecting power is recommended, and it is advised that operation should occur under supervision.

6. Operation instructions

To operate LUMIN, the Arduino is connected to a PC via USB cable to receive power, and allow for the code to be uploaded. Then, each of the three 24 V power supplies is powered on to supply the drivers and motors. The user stands or sits at eye level with the center of the board and presses an LED when it becomes illuminated. Initially, the system runs a calibration to detect the user-specific workspace. For example, the pie slices are moved into their “smallest position,” and then five lights are illuminated across the four slices. When the LEDs are clicked, the device expands, and five more random lights are illuminated. The device continues this until they click all 5 illuminated LEDs. This is used to determine the maximum expansion size of the pie slices. During regular operation, sequences of illuminated LEDs, pie slice movement, rotation angles, rotation speeds, and other parameters can be programmed by the user to generate unique tasks. Regardless, all tasks involve the user clicking on the illuminated LEDs.

7. Validation and characterization

LUMIN is intended to be able to be a more accessible, cheaper, lighter, and smaller option compared to what is currently available on the market. The device is $900 which is significantly less expensive than existing systems. The device weighs 38 lbs., which is under the 51 lbs. maximum lifting load set by NIOSH [18]. The minimum device size is 37″ and can be expanded to 64″. As a comparison, the DynaVision is 48″ x 48″ in size. Finally, LUMIN does not need to be attached to a wall, put on a large or heavy stand, as it just needs to be placed on a table suited for the patient’s sitting or standing height.

LUMIN can extend its pie slice out to 32″ from the center, giving patients a workspace area of 64″ diameter workspace. While at its minimum position, the pie slice is 18.5″ from the center, providing a 37″ diameter workspace. When the device is in the minimum position, there are four buttons 5″ from the center for short reach and interior reaching tasks. These dimensions are compatible within 98.5% of the average female’s wingspan and 90% of the average male’s wingspan. In comparison, the DynaVision provides 73% and 67% of the average female’s and male’s wingspan, respectively [19].

The operational linear speed of the Pie Slice and Angular Velocity of the control arms was 1.7 cm/s and 54.2°/s, respectively (eq.X and Y).

$$\mathit{Linear}Speed(V_l)=\frac{\mathit{Lengt}{h}_{\mathit{start}}(in)-Lengt{h}_{\mathit{end}}(in)}{\mathit{Time}(sec)}=\frac{4\frac{1}{4}(in)-\frac{1}{4}(in)}{3.0sec}=0.67\frac{\mathit{in}}{\mathit{sec}}=1.7\frac{\mathit{cm}}{\mathit{sec}}$$ (6)

$$\mathit{Angular}Velocity({\omega }_v)=\frac{\mathrm{\Delta }\theta }{\mathrm{\Delta }t}=\frac{{90}^{°}deg-0^{°}deg}{26.96sec-25.30sec}=\frac{{90}^{°}deg}{1.66sec}=54.2\frac{\mathit{deg}}{\mathit{sec}}$$ (7)

The data were collected by recording a video of LUMIN while it was performing a sequence of movements. For the linear movement analysis, the pie slices started at their minimum position, and then they expand outward. For the angular movement, we analyzed the movement with the device rotating clockwise and then rotating back counterclockwise. Each rotation was executed with a 100 degree turn and then a visual measurement of the turn using video was made. For example, time stamps were recorded when the arm on the device was at the defined 0-degree position (vertical) and again at 90 degrees. The time difference was recorded, and the average angular velocity was recorded. These performance metrics support a broad range of experiments including pathological movement assessments, and athletic performance assessments [20].

To test LED latency, four-time stamps were taken from the same video. The time stamps occur when the light turns on and then off for two cycles. Below is the calculation for LED latency.

$$\mathit{Total}Timeon(t_{\mathit{total}on1})=|LightOff(t_{off1})-LightOn(t_{on1})|=3.00s-2.47s=0.53s$$ (8)

$$\mathit{Total}Timeon(t_{\mathit{total}on2})=|LightOff(t_{off2})-LightOn(t_{on2})|=3.70s-3.20s=0.50s$$ (9)

$$\mathit{Average}TimeOn=\frac{t_{\mathit{total}on1}+t_{\mathit{total}on2}}{2}=\frac{0.53+0.50}{2}=0.515s$$ (10)

$$\mathit{Latency}=|Expectedtimeon(t_{\mathit{total}onexp})-Actualtimeon(t_{\mathit{total}onact})|=|0.5s-0.515s|=0.015s$$ (11)

$$\mathit{Total}Timeoff(t_{\mathit{total}off})=|LightOn(t_{\mathit{on}})-LightOff(t_{\mathit{off}})|=3.20s-3.00s=0.2s$$ (12)

$$\mathit{Latenc}{y}_{\mathit{off}}=|Expectedtimeoff(t_{\mathit{total}offexp})-Actualtimeoff(t_{\mathit{total}offact})|=|0.2s-0.2s|=0s$$ (13)

The code states that the light is to turn on then have a 500 ms delay before turning off and have a 200 ms delay before turning on again. The is ∼ 0.015 s of latency when turning light on to off and ∼ 0 s of latency when turning from off to on. There is a factor of human error in the calculations as the time stamps were found manually.

Case study.

An experiment was carried out demonstration operational performance. Data were collected during experimentation to quantify the results. The experiment was performed with one participant who was positioned 1 foot away from the device interface, while the rotational center point of the device was positioned to be level with the participant’s center of wingspan (Sternum). The aim of the study was to gather baseline reaction time data for both the dominant and non-dominant hand as it varies depending on max wingspans and device configuration. The subject started the test by selecting a blinking light in the center of the device (go signal) which started the reaction time clock. With an initial device position at 0 degrees (default starting rotation position of the LUMIN) four lights were randomly displayed in sequence to the user for them to select (i.e., when one light illuminates, the user selects its, and then the next light illuminates). Following the completion of each sequence, the LUMIN rotated by 22.5 degrees where a total trial thus included positions of 0, 22.5, 45, and 67.5 degrees. In total, twenty lights displayed and selected by the user per trial. After each rotation the device would simultaneously flash all the lights on the LUMIN as an indicator that the reaction timer was restarted. Each trial was then repeated at different “max wing span” LUMIN configurations (i.e., increasing the diameter of the board) – 32”, 35.5”, 39”, 42.5” and 46”.

The results are shown in Fig. 8 with dominant (blue) and non-dominant (orange) reaction time plotted with respect to wing span (Fig. 8a) and rotational position (Fig. 8b). As expect, task time (i.e., time between when the light turns on an when it is selected) is longer for both hands as the light is further away from the center-point. Specifically, the mean reaction times are 0.530 s (dominant) and 0.513 s (non-dominant) at the 32″ position, and 0.702 s (dominant) and 0.664 s (non-dominant) at the 46″ position. Indicating an increase of 0.172 and 0.150 s (32.5% and 29.3%) for the dominant and non-dominant hand, respectively. For the reaction-time vs rotational position plot (Fig. 8b), the times are largely equivalent at each position. For example, the average dominant hand reaction-time was 0.594 ± 0.157 s, while the non-dominant hand was 0.592 ± 0.142 s. Generally, it is expected that the dominant hand has a shorter reaction time due to habitual use and a higher neural efficiency [21]. However, some variability in task-execution can result in inverted reaction-time results. Regardless, this case study is not representative due to the limited study population but instead demonstrates the utility of the system in experiments to measure reaction-time performance using unique spatial and dynamic configurations.

Fig. 8.

Example data generated from using the LUMIN system for a case study. (a) Reaction-time vs wing-span experiment where the LUMIN board was gradually increased in size; (b) Reaction-time vs. rotational position plot where the LUMIN rotated by 22.5 degrees after sequences of movements. Dominant hand results are shown in blue, while non-dominant hand results are shown in orange.

Ethics statements

The research of this device was focused on the development of the hardware for our future studies; therefore, no human or animal studies were conducted in this work.

CRediT authorship contribution statement

Zachary D. Peterson: Writing – original draft, Validation, Methodology, Investigation, Data curation, Conceptualization. Filip Stefanovic: Writing – review & editing, Supervision.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Biographies

Zachary D Peterson is a MS Research Assistant in the Department of Biomedical Engineering at the University at Buffalo, where he is a part of the Computational Neurorehabilitation and Biomechatronics (COMAND) Lab. His research is focused on Neurorehabilitation devices, CAD modeling and 3D printing, and Mechanical and Electrical Systems.

Dr. Filip Stefanovic is an Assistant Professor of Biomedical Engineering at the University at Buffalo. His research interests include the development of wearable neurorehabilitation technologies, computational neurophysiology in rehabilitation, and high-density sensors. Filip has experience working in industry as an R&D Engineer developing Autonomous IoT systems, predictive AI algorithms, and electromyography medical devices. He received his PhD at McGill University in Montreal, Canada, his Master’s degree at the University of Belgrade, Serbia, in conjunction with Aalborg University, Denmark, and his BASc at the University of Toronto.