FINAL DESIGN

Mechanical

The mechanical design process of the project was guided by two main constraints: replicating accurate human shoulder movement by achieving the full range of motion (ROM) necessary to perform upper arm repositioning (UAR) and ensuring anatomically precise shoulder movements. Specifically, having a singular center of rotation (COR) and reaching the full ROM of a human shoulder were difficult features to find in the same design.

Initially, a Pugh chart was created to evaluate various shoulder joint design ideas based on criteria such as cost, familiarity, mass, range of motion, and others. This evaluation led to the selection of two designs for prototyping. Initial prototyping information can be seen here in greater detail.

Both prototypes were developed and tested, and the final design emerged as a combination of the best elements from each. This hybrid and novel design, however, still required important modifications to optimize its functionality. The main two changes applied were: 

• Inverted the small basket orientation to create an inverted u-joint

  • This enabled the shoulder joint to achieve better weight counterbalance

• Added a wedge to the separate joint (SJ) base

  • This enabled the shoulder joint to access a greater range of motion to accurately replicate the specific motion of upper arm repositioning

These refinements were crucial in ensuring that the final design not only met the project's original constraints but also provided a robust and accurate replication of human shoulder motion. The iterative process of prototyping and refining enabled us to achieve a high-performing mechanical component for advanced physical human robot interaction applications.

Software

The software component of the project can be broken down into two separate functions: sensing position and applying torque feedback. To reduce risk and ensure the function of each feature, an incremental approach to software complexity was utilized. For example, before proceeding with torque feedback, it was important to verify that reliable position data could be obtained from the encoders. Similarly, software was first programmed and tested in a 1DOF system before being implemented onto a 3DOF system to better isolate any potential issues. 

Electronic components were chosen based on project needs and budget constraints. Cheap electronics were initially used to demonstrate functionality and eliminate purchase paralysis to the detriment of hardware performance. This proved successful in helping keep costs low by avoiding over-engineering of components and gaining a better understanding of hardware limitations through rigorous testing.

Sensing + Torque Feedback

The code implemented on the haptic shoulder project is built using C++ for its ease of implementation with the Arduino IDE framework and its ability to do object-oriented programming. Each motor is initialized with a motor identity, encoder pins, motor pins, range-of-motion limits, and dynamic parameters (stiffness constant and damper ratio). The scripts are stored at this repository.

To simulate the resistive forces of a shoulder, a virtual spring and damper were modeled in software based on the position of the upper arm.

PERFORMANCE

Range of Motion (ROM) Analysis

The only movement for which the SHULDRD does not access the same ROM as the human shoulder is flexion. To perform UAR, however, flexion is not executed while the extension motion is. Therefore, a trade-off between losing flexion ROM and gaining extension ROM was necessary to fulfill the project's objective. 

Electronic Components

Torque Feedback Flowchart

SHULDR Device State Machine

Software Class Diagram