Drone Speed Calculator

I’ve revisited the drone sidecar .srt script to make the output even simpler. It now generates an .srt file specifically for importing into Premiere, displaying only speed in miles per hour. This is achieved using the haversine formula, which calculates the distance between two points on a sphere. Here’s a breakdown of the script and its applications.

My primary drone is the Mavic 3 Classic, which features three modes on the top of the controller: Cinematic, Normal, and Sport.

For motion aerial imagery, I typically use Cinematic mode for low-altitude shots and Normal mode for higher-altitude shots. The drone is limited to 400 feet above ground level in most cases, and I rarely use Sport mode unless I need to cover long distances quickly while flying near the maximum relative altitude.

The drone relies on multiple global navigation satellite systems (GNSS) for positioning, including GPS (Global Positioning System) from the USA. According to ChatGPT, "the Mavic 3 Classic typically connects to 10–30 satellites at a time for accurate positioning."

GNSS Accuracy:

• Horizontal positioning accuracy: ±1.5 meters (4.9 feet)
• Altitude accuracy: ±0.5 meters (1.6 feet)

The speed calculations in my script use a rolling 48-frame window, which provides a 2-second interval (at 24 fps) for normalizing the data points. This improves the accuracy of the calculation by reducing the impact of outliers that could skew the results. Without this rolling window, the data might produce unreliable speed outputs.

I did the calculation by hand to appreciate the results, comparing the delta between Frame 1 and Frame 1264. In clip below, the drone flies 0.35 miles at an average speed of 24 MPH. The Python script, which calculates speed every 48 frames (2 seconds), performed this operation 26 times—in a fraction of a second—to generate the simplified .srt displayed at the bottom of the frame. By comparison, it took me 45 minutes to do a single calculation by hand with pen and paper.