Tracer - A Wearable for your Things!

Have you ever wanted to just stick an IMU on a thing to trace its pose and motion? Tracer provides reliable high-performance and cost-efficient tracking of objects for a variety of applications. Tracer can be easily mounted using a high-quality velcro strap, allowing it to be fastened on the frame of a bicycle for lean and cadence tracking or even the neck of a tennis racquet to count several strokes and potentially differentiate between topspin and slice. Let your imagination run wild!

Before diving further, I have to thank JLCPCB for sponsoring this project. They have been a crucial part of the process by enabling me rapid prototype without any compromises in quality. All of the PCBAs related to this project were prototyped and assembled by them. I highly recommend them as they are high quality, cost, and time-efficient, if you do intend to design and build any boards of your own.

Leveraging on existing open-source projects the Tracer fully supports integration with Phyphox and can be programmed using Micropython, and Arduino. Power users can also use the ESP-IDF tool for maximum customization.

Key Features;

• Leverage the LSM6DSL to track your objects in real-time.
• Easy mounting using high-quality velcro straps that allow the unit to be mounted on a wide variety of objects from your bicycle to track lean angles to a tennis racquet for pose insights.
• On-board Li-ion battery charging using the TP4065.
• Battery Life - 3 hours of continuous streaming over BLE at 10Hz.
• 15m BLE range (unobstructed) - tested on a tennis court

Available to buy on Tindie.

Specification

Microcontroller | ESP32-PICO-D4

• WiFi | 802.11b/g/n
• Bluetooth | BLE 4.2
• FLASH | 4MB
• Programming | USB over UART (CP2104)

Inertial Measurement Unit (IMU) | LSM6DSLTR

• Accelerometer | ±2/±4/±8/±16 g at 1.6 Hz to 6.7KHz
• Gyroscope | ±125/±250/±500/±1000/±2000 DPS at 12.5 Hz to 6.7 kHz

Time-of-Flight (TOF) | VL53L0X (Swappable to VL53L1X)

• Range | Up to 2m
• 1D Gesture Recognition

Li-ion Battery Management | TP4065

Power | 700mA 3.3V LDO

Mechanical

• Weight | 20g, including velcro strap
• Dimensions | 4.2 x 3.6 cm
• Mounting Options | Velcro Strap or 3x M.2.Screws

Use Cases

The example code used to test this design involves close integration with the Madgwick Filter. Using that filter to perform sensor fusion and generate an estimate for heading, roll, and pitch, the data can then be relayed to a smartphone or PC for post-processing. See below for two example

Real-time 3D Visualisation

Real-time 3D Visualisation

This 3D visualization is by streaming data over UART to a simple script that was written on [Processing](https://github.com/processing).

Stream real-time data to Phyphox

The real-time plots from the IMU are streamed over BLE to the phone. Data can later be exported over CSV for further analysis if required. This example shows the Tracer streaming the accelerometer data and no. of tennis ball strikes over Phyphox.

Real-time gesture recognition using Edge Impulse

This example illustrates the Tracer's ability to be highly customizable to suit a wide variety of applications and specifications. Using the Edge Impulse platform, I was able to train a neural network that would then run on the ESP32 natively to track a certain gesture or activity. More details on this here.