analysis

Telemetry Extraction & Analysis

The Centauri quadcopter logs telemetry directly onto the Raspberry Pi Pico's onboard flash storage, several times per second. After a flight, you can connect the Pico to a computer and extract the telemetry file (log in the root directory). This file is stored in a compact binary format, with densely encoded telemetry packets. To interpret the data, you'll need a separate script that unpacks it into a human‑readable format such as a .csv.

This analysis script performs that conversion. When run, it:

1. Prompts for the path to the binary log file on your computer, then unpacks all the telemetry frames within it.
2. Infers each individual "flight" in the log (a span of time when the quadcopter went from unarmed --> armed --> unarmed again) and prints summary stats for each: duration, battery voltage range, G-force range, average throttle, and Rx (command receipt) latency.
3. Prompts for a target path to save the unpacked telemetry to as a .csv file.

As you can see below, the Centauri quadcopter will log telemetry as a binary file named log. That data can be retrieved from the microcontroller via Thonny.

Use can use this analysis script to decode the binary log file and generate a structured .csv file for easy inspection. Below is an example of the telemetry output:

As shown above, the telemetry maps multiple onboard parameters against a timestamp, including:

• Seconds - the exact timestamp, in seconds, this telemetry frame was taken at (note this may decrease if the quadcopter is rebooted, resetting system time back to 0)
• Battery Voltage - supply voltage from the onboard battery
• Pitch Rate - actual pitch rate (°/s)
• Roll Rate - actual roll rate (°/s)
• Yaw Rate - actual yaw rate (°/s)
• Pitch Angle and Roll Angle - angle estimates calculated by the flight controller's complementary filter
• G-Force - how many G's the quadcopter is pulling at that moment
• Input Throttle - pilot throttle input, expressed as a percentage (0–100)
• Input Pitch - pilot pitch input, expressed as a percentage (−100 to 100)
• Input Roll - pilot roll input, expressed as a percentage (−100 to 100)
• Input Yaw - pilot yaw input, expressed as a percentage (−100 to 100)
• M1 Throttle - front‑left motor throttle (calculated by the flight controller)
• M2 Throttle - front‑right motor throttle (calculated by the flight controller)
• M3 Throttle - rear‑left motor throttle (calculated by the flight controller)
• M4 Throttle - rear‑right motor throttle (calculated by the flight controller)
• Cmd Last Received ms - how long ago (in milliseconds) a Control Packet was last received from the transmitter, helping indicate Rx issues

Try it yourself!

You can try converting and analyzing Centauri telemetry with nothing more than a computer that has Python installed. Below are a few direct download links to raw log binary files to experiment with:

• February 13, 2026: Corrupt Packet Failure
• February 22, 2026: eight flights
• March 7, 2026: multiple multi-minute flights

Download a log file above and pass it to main.py to unpack it, save it as a CSV file, and test your analysis skills!

Overlaying Telemetry on Video & Dashboarding

Beyond just inspecting the raw numbers, the unpacked telemetry can also be used to overlay live flight data onto video recordings, or fed into any dashboarding/reporting tool (Power BI, Tableau, or even an AI-generated dashboard). I built Centauri-specific Dashware assets that let you overlay telemetry data directly onto a video with only a few clicks and timing syncs.