MCU Integration Research #3
Description
Introduction
In this year's exoskeleton design we are hoping to integrate the Raspberry Pi with multiple STM32 microcontrollers. The goal is to have a microcontroller at each joint, which interfaces with the IMU and Motor to have a quick feedback loop with the control system's output tuning.
We will also pull the IMU and motor encoder sensor readings onto the Raspberry Pi to use in our heavier machine learning algorithms to send less frequent predictive positioning commands to the motors.
Goal
Determine the feasibility of having a micro controller at each joint of the exoskeleton, exploring different design approaches (2 micro controllers on each leg?) and determining the most optimal communication methods to transfer data between the Raspberry Pi and STM32 boards.
Requirements
Consider the following metrics while weighing the feasibility of each MCU implementation:
- Cost
- Per MCU ~ $40, for 6 ~ $240
- Processing Speed
- Data for AI training samples every 0.05 seconds (20hz) from sensors and motors
- I/O capabilities / Communication
- Ease of implementation
- Error handling / Monitoring
Software Members should also go into more detail on the following in their research:
- Validation of STM32 boards selected by electrical / estimated processing requirements for control loop.
- Communication implementation (MCU to PI, IMU to MCU, Motor to MCU)
Deliverables
Each assignee is required to:
- Using the requirements stated above write a comprehensive performance comparison of the two different strategies (1 MCU per joint vs 1 MCU per leg). This should be documented in here (Your work should go under your designated document tab).
- Visualize this data with a diagram, drawing, chart or any other means for implementation. This will be helpful to visualize the implementation and give mechanical an idea of what we're looking for.
- Leave a comment on this issue indicating the completion of your research with a brief conclusion based on findings.