Project

# Title Team Members TA Documents Sponsor
52 Digital Coaching for Figure Skating
Ethan Yee
Lionel Binder
Stephanie Tancs
Zhicong Fan design_document3.pdf
final_paper1.pdf
photo1.png
presentation1.pptx
proposal2.pdf
# DIGITAL COACHING FOR FIGURE SKATING

Team Members:
- Stephanie Tancs (stancs2)
- Lionel Binder (lbinder2)
- Ethan Yee (ethanay2)

# Problem

Coaching for figure skating costs $50+ / 25 min of instruction. With the costs of ice time and skate maintenance outside of this, the cost of figure skating is significant if you are looking to improve your skills in any way. This makes it impossible for many who would like to improve or begin figure skating to do so based solely on cost. For two years of twice-weekly coaching, the cost would be $10,400 for coaching alone.

# Solution

With a system consisting of wearable electronics, a camera setup, and access to a computer, each ice rink could be equipped for digital coaching and encourage skaters to improve without needing to invest significant amounts of money into coaching. The system would function as the wearable electronics can record accelerations in 3 dimensions, and the software can synthesize this information and the input from the camera in order to create a model of the skater and compare it directly with an ideal model. This is so the skater can directly see what they are attempting and make specific changes to their motion in order to perfect their form. Furthermore, this solution would aid the coaching of those who have a different style of learning (visual, tactile) as opposed to auditory to understand the corrections made.

# Solution Components

## IMU nodes

We will manufacture multiple IMU nodes consisting of an accelerometer and gyroscope to be outfitted on an individual at important body joints. These nodes will be able to record acceleration data from the wearer and subsequently transmit through a wired connection to the master node that is collecting data. The IMUs make up one half of the wearable electronics. In order to attach these nodes to the skaters, we will CAD and 3D print a protective case, and pin the electronics onto the skater via a safety pin to ensure maintenance of a precise location. The node on the skate will be attached via a safety pin to a sleeve (commonly sold as a skate protector to freestyle skaters) that goes over the boot of the skate.

## Microcontroller

We will also manufacture a custom microcontroller that serves as the master node for the wearable electronics system. This microcontroller will need to be wired to all of the IMU nodes in order to receive and aggregate the wearer’s acceleration data. It will also require a storage system, likely SD card, in order to later transfer the accumulated data to the software component of the project. We will power the entire wearable system through the microcontroller. We also want to keep in mind that we find compatible IMUs which operate on the same communication protocol as the microcontroller. The microcontroller makes up the second half of the wearable electronics.

## Camera and Software

We will need to connect a camera to a computer in order to accomplish the computer vision aspect of this project. The camera will capture the motion of the skater as well as track the world position of the IMU nodes to provide visual data. This will be paired with the IMU data to generate a model of the skater’s movement as they execute skill moves. We can compare these models with “ideal” models based on skilled figure skaters to both quantify the difference between a performed move versus an “ideal” move and to provide visual feedback to the skater as to how they can improve their performance. We plan to use OpenCV or some similar library to accomplish this.

# Criterion For Success

This can be demonstrated in the UI Ice Arena with demonstrators from the club synchronized figure skating team. This demonstration can be easily video recorded and the data can be processed and the program can be run in class.
Our high-level goals are as follows:
- Design and build a wearable electronics system to measure the acceleration data of an ice skater performing the Biellmann (a skating skill in which the skater lifts the leg above their head and grabs onto the blade with their hands)
- Utilizing camera input to align with acceleration data and create a full depiction of the skater in terms of kinematics
- Use aggregated data to generate a visualization of the skater and quantify the difference between an “ideal” move versus the skater’s move

Resonant Cavity Field Profiler

Salaj Ganesh, Max Goin, Furkan Yazici

Resonant Cavity Field Profiler

Featured Project

# Team Members:

- Max Goin (jgoin2)

- Furkan Yazici (fyazici2)

- Salaj Ganesh (salajg2)

# Problem

We are interested in completing the project proposal submitted by Starfire for designing a device to tune Resonant Cavity Particle Accelerators. We are working with Tom Houlahan, the engineer responsible for the project, and have met with him to discuss the project already.

Resonant Cavity Particle Accelerators require fine control and characterization of their electric field to function correctly. This can be accomplished by pulling a metal bead through the cavities displacing empty volume occupied by the field, resulting in measurable changes to its operation. This is typically done manually, which is very time-consuming (can take up to 2 days).

# Solution

We intend on massively speeding up this process by designing an apparatus to automate the process using a microcontroller and stepper motor driver. This device will move the bead through all 4 cavities of the accelerator while simultaneously making measurements to estimate the current field conditions in response to the bead. This will help technicians properly tune the cavities to obtain optimum performance.

# Solution Components

## MCU:

STM32Fxxx (depending on availability)

Supplies drive signals to a stepper motor to step the metal bead through the 4 quadrants of the RF cavity. Controls a front panel to indicate the current state of the system. Communicates to an external computer to allow the user to set operating conditions and to log position and field intensity data for further analysis.

An MCU with a decent onboard ADC and DAC would be preferred to keep design complexity minimum. Otherwise, high MIPS performance isn’t critical.

## Frequency-Lock Circuitry:

Maintains a drive frequency that is equal to the resonant frequency. A series of op-amps will filter and form a control loop from output signals from the RF front end before sampling by the ADCs. 2 Op-Amps will be required for this task with no specific performance requirements.

## AC/DC Conversion & Regulation:

Takes an AC voltage(120V, 60Hz) from the wall and supplies a stable DC voltage to power MCU and motor driver. Ripple output must meet minimum specifications as stated in the selected MCU datasheet.

## Stepper Drive:

IC to control a stepper motor. There are many options available, for example, a Trinamic TMC2100. Any stepper driver with a decent resolution will work just fine. The stepper motor will not experience large loading, so the part choice can be very flexible.

## ADC/DAC:

Samples feedback signals from the RF front end and outputs the digital signal to MCU. This component may also be built into the MCU.

## Front Panel Indicator:

Displays the system's current state, most likely a couple of LEDs indicating progress/completion of tuning.

## USB Interface:

Establishes communication between the MCU and computer. This component may also be built into the MCU.

## Software:

Logs the data gathered by the MCU for future use over the USB connection. The position of the metal ball and phase shift will be recorded for analysis.

## Test Bed:

We will have a small (~ 1 foot) proof of concept accelerator for the purposes of testing. It will be supplied by Starfire with the required hardware for testing. This can be left in the lab for us to use as needed. The final demonstration will be with a full-size accelerator.

# Criterion For Success:

- Demonstrate successful field characterization within the resonant cavities on a full-sized accelerator.

- Data will be logged on a PC for later use.

- Characterization completion will be faster than current methods.

- The device would not need any input from an operator until completion.

Project Videos