Abstract
Our previous IoT based prosthetic arm prototype used servo motors to control finger movement through an Arduino Mega, which is connected to the muscle, pulse, and temperature sensors. The Arduino Mega was also connected to a Raspberry Pi 3 model B to transfer data from/to an online web application. One major limitation encountered during testing this prosthetic device was the space occupied by these components, which makes the device bulky. In addition, these servo motors cannot control the movement of the prosthetic device precisely. In this paper, we propose to improve on the existing prosthetic limb prototype by transitioning the electromechanical system to linear actuators and replacing the larger Arduino Mega and Raspberry Pi 3 with the smaller Arduino Nano and Raspberry Pi Zero W, respectively. These changes will result in a more cost-effective, more stable, and more accurate prototype, resulting in better performance.
Introduction
Developing functional prosthetic limbs is a complex balancing act between high-level capability and daily wearability. Previous iterations of the "Myo-prosthetic" arm developed at the New York City College of Technology faced two primary hurdles that hindered real-world application. First, the bulk and weight of standard-sized components, such as the Arduino Mega and Raspberry Pi 3, made the device cumbersome for users. Second, the reliance on servo motors and string-based finger control lacked the necessary precision for delicate tasks and suffered from high power consumption to maintain a grip. This research aims to transition the prototype from a proof-of-concept to a functional medical device by optimizing the hardware footprint and mechanical drive system.
Methods and System Architecture
The researchers adopted a "divide and conquer" architecture, splitting the system into two distinct subsystems to optimize performance and space.
Motion Subsystem
An Arduino Nano was selected to control five linear actuators—one for each finger—via A4988 stepper motor drivers. This represents a significant shift from hobbyist servos. These actuators utilize a screw-and-slider mechanism, translating the circular motion of a stepper motor into precise linear displacement.
Sensor & IoT Subsystem
A second Arduino Nano handles data collection from a suite of biosensors, including Electromyography (EMG) for muscle activity, pulse sensors, and temperature probes. This data is relayed to a Raspberry Pi Zero W, which serves as an IoT gateway. The Pi Zero W uploads health metrics to a Firebase cloud database, enabling remote monitoring by healthcare professionals.
Empirical Calibration
A critical component of the methodology was the empirical calibration of the actuators. Recognizing that manufacturer-provided specifications can be unreliable in real-world mechanical assemblies, the team developed custom code to measure actual displacement. By testing across a range of 200 to 800 pulses, they established a precise "pulse-to-millimeter" ratio, correcting for mechanical variances.
Key Insights and Results
The redesign yielded several significant improvements in efficiency, size, and control.
Mechanical Efficiency
The move to screw-slider linear actuators introduced a "holding" advantage. Due to the mechanical friction inherent in the screw thread, the fingers can maintain their position even when power is disconnected. This drastically reduces energy consumption compared to servo-based designs, which require constant current to hold a grip.
Footprint Reduction
The hardware optimization achieved massive space savings. The switch to the Arduino Nano resulted in a controller footprint 85% smaller than the previous Arduino Mega. Similarly, the Raspberry Pi Zero W provided a 50% size reduction over the Pi 3 Model B, significantly improving the device's wearability.
Precision and Power Management
Empirical testing determined that 67 pulses correspond to exactly 1mm of movement, allowing for highly granular control. Furthermore, the system successfully implemented dual power rails (9V for motors and 5V for logic) to ensure that electrical noise from the actuators does not interfere with sensitive biosensor data.
Conclusion and Future Work
While the internal electronics and mechanical drive have been successfully optimized, challenges remain regarding the overall weight and user interface. Future research will focus on optimizing the 3D printing process (Fused Deposition Modeling) and exploring alternative materials to further reduce weight. Additionally, the team plans to refine the physical socket design to improve long-term user comfort and investigate advanced 3D printing techniques to enhance both the structural integrity and aesthetic appeal of the prosthetic shell.
Summary updated