quadcopter-telemetry-genie

# Quadcopter flight telemetry: deploy and explore with Genie The bundle in this folder loads a static release (`data/*.parquet`) into Unity Catalog and creates a Genie space over the raw tables. Read [`README.md`](README.md) for the story and [`data/README.md`](data/README.md) for every column, frame and sign before reasoning about the data. ## 1. Check the prerequisites - `databricks --version` is 1.3.0 or later (the `genie_spaces` resource and the direct engine need it), and the CLI is authenticated to the target workspace. - Ask the user for an existing catalog where they can create a schema, and a Pro or serverless SQL warehouse id (`databricks warehouses list` shows candidates). Never guess either. ## 2. Deploy in three steps From this folder: ```bash export BUNDLE_VAR_catalog=<catalog> BUNDLE_VAR_warehouse_id=<warehouse_id> databricks bundle validate databricks bundle deploy # first time: expected to fail at the Genie space ("Table ... does not exist") databricks bundle run load_release0 # writes flights, telemetry, landings, blended_wind_reference databricks bundle deploy # creates the Genie space ``` - The first deploy's failure at the Genie space is expected: the space needs its tables, and only the job creates them. Anything else failing in that deploy is a real error. - Check the load: the job's notebook output (or `SELECT COUNT(*)` on each table) must give flights 13, telemetry 33,851, landings 6, blended_wind_reference 33,851. ## 3. Explore with Genie Open Genie, pick *Quadcopter Flight Telemetry*, and ask each question in a fresh conversation. Compare with the reference answers; Genie writes its own SQL, so small differences are normal. | Question | Reference | |---|---| | Which flight was the gustiest? Use the biggest short gust above the flight's average wind. | The `day_2` automated-manoeuvres flight: 3 s gust 1.75 m/s above a 3.00 m/s mean | | How many of the dock landings seated? | 4 of 6 | | The flight controller only gives wind direction in 45° steps. Using the drone's tilt during steady flight, estimate a finer wind direction for each flight. | Windiest flights: from about 13° to 23°, between the N and NE sectors | Rules when answering from these tables: - `blended_wind_reference` is a reference answer and is not in the Genie space on purpose. Use it only to compare a derived direction, in the SQL editor. - The wind estimate is meaningful only clear of the ground: ignore it on the ground, at take-off, at touchdown and at hand catches, and treat it as noise for the dock flights (they hover under about 1 m). - Wind directions are where the wind blows *from*, clockwise from north, and wrap at 360: average them as vectors, never arithmetically. For the full tilt derivation, step by step with the SQL Genie should reach, follow [`guide/DERIVE_BLENDED_WIND.md`](guide/DERIVE_BLENDED_WIND.md). ## 4. Tear down ```bash databricks bundle destroy ``` If the schema delete is refused because it still holds tables, run `DROP SCHEMA <catalog>.quadcopter_flights CASCADE` and then `destroy` again. Confirm with the user before dropping anything.