Happy Friday! Throttle control is working! (Patreon)

Hey Patrons, REALLY quick update here. I've got throttle control working!

Lemme give you a sense at what you're looking at. All the plots are vertical position, vertical velocity, and body axis thrust(the throttling isn't represented in the thrust here yet). In a few plots you can see that when the vehicle hits the ground it bounces back up. You can also see drag as a small gray line, as well as the vehicle orientation in the throttling plots. The cool part, is that this is all running on AVA, not a fancy offline sim. AVA is drawing out these plots. Each sim takes about 4-6ms, so we'll be running multiple of them during the actual flight to hone in on a safe landing altitude as we near the end of the burn.

I'm randomizing the ignition time of the retro rocket motor, just like we see in real life. That's the main trouble with landing, getting the motor to start exactly when you want. In the first plots, you can see the rocket just can't land. The timings are about 30-40% greater than we can expect to see in flight, and some of those landings are quite low/high.

During the "throttled" simulations, the vehicle measures the error in altitude after the burn has begun, then angles itself along a sine wave. The period of the wave is fixed at 2 seconds, but the magnitude is variable, which is how we control how much energy is spent in the vertical direction. I typed the following out in Discord last night, but I'll drop it here as well. I'm describing the process as it should happen in flight, with the spike being the thrust spike at the start of the burn.

"so we hit the gas, detect the spike, then sim the rest of the flight(what's happening here). We use the error in our altitude to say "oops we're a little high, we need to bleed off 7%". We use that 7% to generate a sine wave with a fixed period but a variable magnitude. Ultimately, we'll do this multiple times as we get closer, but after the wave completes, we leave the last 0.5 seconds of the burn for a cranked up control gains section. That's our settling maneuver."

There's more work to do here, as we only have one throttle point. I'd like to inject a ton of noise in here to see how things play out then, and the actual math that gets us the bleedoff percentage needs work too. One step at a time, but this is a fun way to close the week!

Blue skies,

Joe