multi-player and flightaxis?
- Thread starter tridge
- Start date
The multiplayer feature hasn't been documented to my knowledge. We don't have any plans to create such documentation, I'm afraid.
That said, there should be nothing to stop you from networking two (or more) FlightAxis sessions together into a multiplayer session. The vehicles could then share a reality, and the controllers can talk among themselves using whatever protocol you'd use in the world.
In the case of machine vision, you'd use the onboard cameras using your existing solutions.
It's not quite as simple as injecting a synthetic vehicle via multiplayer, but it does add extra verisimilitude!
Let us know if you get this going - I've always wanted to try it.
That said, there should be nothing to stop you from networking two (or more) FlightAxis sessions together into a multiplayer session. The vehicles could then share a reality, and the controllers can talk among themselves using whatever protocol you'd use in the world.
In the case of machine vision, you'd use the onboard cameras using your existing solutions.
It's not quite as simple as injecting a synthetic vehicle via multiplayer, but it does add extra verisimilitude!
Let us know if you get this going - I've always wanted to try it.
tridge
Member
The project that led to this question was one where we are adding a feature to ArduPilot quadplanes where the quadplane can takeoff and land on a moving ship. I've added support for that in ArduPilot, and added simulation of the ship movement in SITL (software in the loop simulation).
So what I was imagining for bringing this to RF9 was:
It would be nice to know if the above is at least theoretically possible before I put a lot of time into working out the protocol. For example, if the protocol included license key info that I can't generate in SITL then it wouldn't be worth trying.
Cheers, Tridge
So what I was imagining for bringing this to RF9 was:
- build a model of a ship with a heli-pad in 3ds max
- import as a model in RF9. The model wouldn't actually need any method of propulsion, it would just be a ship that sits on the water
- make ArduPilot SITL pretend to be a multi-player client in RF9, and send the main RF9 instance data telling it that there is a player with a ship at the following (moving) position and pose
It would be nice to know if the above is at least theoretically possible before I put a lot of time into working out the protocol. For example, if the protocol included license key info that I can't generate in SITL then it wouldn't be worth trying.
Cheers, Tridge
tridge
Member
we've been playing with this, and we're finding the aircraft just pass through each other. We've been trying to land a quad on the back of a hovering quadplane.
For the protocol, it looks like it is RakNet: https://github.com/facebookarchive/RakNet
For the protocol, it looks like it is RakNet: https://github.com/facebookarchive/RakNet
uncle twist
Well-known member
Speaking to the aircraft going transparent in multiplayer...THAT is what happens when an aircraft sits for about 10-15 seconds...??? in multiplayer without any input from the controller (you/TX).... I`m pretty sure it doesn`t matter where you are on the field. I think it was designed that way in order for planes not to crash/interfere with each other upon spawning and or interning that particular session of multiplayer. I`m also wondering if/is it possible to give the landing platform the physics/hardness of a runway so the quads skids don`t sink into it. Just thinking out loud tridge, very cool project you`ve got going on... ? 
tridge
Member
I noticed that when on the ground, but with ArduPilot in control there is pilot input at 300Hz. The fact the aircraft keeps hovering also means the motors are still being controlled (quadplanes don't hover for long without changes in motor demands).
So I think there is some secondary condition for this happening, and it seems related to having a 2nd aircraft in close proximity (maybe amount of overlap of aircraft?).
I tried upping the strength multiplier on the landing platform to 1500% (the max). No luck yet. I've never tried creating a runway to know how you setup the physics for it.
I've been head-down on a couple of projects, so it looks like I've got a good backlog of questions to answer here! Let's dig in!
I do think that Multimode is much more likely to succeed than multiplayer for actual contact. I'm still not sure if the resting contact will hold up to a moving "runway". Can Mark test this question to make sure the physics itself can hold up?
This is very neat stuff!
PS: I really appreciate our new forum system's quoting abilities.
I have some concerns about the aircraft successfully resting on a moving vehicle. I would expect it to make contact and receive impulses. But it may end up jumping around a lot, particularly with the discontinuous motion that tends to arise from networked multiplayer.
Interesting. I wonder why collision isn't working (when not ghosted). I know that we experience mid-airs on a fairly regular basis when doing multiplayer testing. Maybe that changes when the contact is very slow. There are a lot of provisions to prevent collision when spawning in at common starting points. I'll have to investigate a bit to comment more knowledgably.
This part is just FYI: The strength multiplier only controls the propensity of a part to break off. Setting it to 1500% means that the part will never break off. That's probably a good idea here.
I'm not sure. Guessing without digging into code, this will either be very possible or nearly impossible. I'll have to investigate before giving any encouraging words.
I do think that Multimode is much more likely to succeed than multiplayer for actual contact. I'm still not sure if the resting contact will hold up to a moving "runway". Can Mark test this question to make sure the physics itself can hold up?
This is very neat stuff!
PS: I really appreciate our new forum system's quoting abilities.