Project Malmo is a platform for Artificial Intelligence experimentation and research built on top of Minecraft. We aim to inspire a new generation of research into challenging new problems presented by this unique environment.
Hi, does anyone know if there is any way to get all the GUI/Inventory stuff to show up as well from a VideoProducer AgentHandler? thx ;D
Hey! I'm new to project malmo(I'm a CS student) and I wanted to find out if I can implement my own deep neural network for it to learn how to do stuff(gathering wood, mining blocks etc) but I didn't understand how to actually put my python code into it.(I want to implement my own network and define it's layers/optimizers/loss function etc) Is there any way to do it?
@qwr1000 did you do the included tutorial? because if so you can just copy one of the python scripts, from tutorial 1 or 2 and implement your own code. You can make other scripts with classes and use tensorflow as you would normally and call functions of your tensorflow script to train and get results from your network inside the modified tutorial script ;D
Hey again! I got it working which it awesome, but I've found another problem. I want to spawn the player with a specific thing in the inventory but I don't know where to get the item names(I know theres "diamond_pickaxe" and "rabbit_stew") are they the normal MC item names, or is there a list of items that Malmo implemented?
Anyone have any good links for creating an online server using malmo. I'd like to host a server and have users play with an agent I designed to see if it is helpful to them.
malmö runs on an older version of minecraft, so if you managed to pull that off, then the clients would need to degrade their minecraft
@rohanpaleja27 the main problem I would think is that the malmö mod is a client mod without server version, so i'm pretty sure you can't make a server with the mod, meaning you can't join if you have the mod installed...
Rohan R Paleja
ah I see, thanks @airzocker_gitlab@marcus1337
Hey, anyone has a quick idea about how to calculate a distance between a block and a player?
if you have two points in space you can take point A minus point B
This might be a somewhat basic question, but: is there a canonical way to use the information returned by ObservationFromGrid, along with the Pitch and Yaw of an agent's viewpoint, to subset the grid observation to those blocks an agent could see from their viewpoint? My understanding is that ObservationFromGrid is returned in absolute orientation rather than orientation with reference to the player's viewpoint, but I can't find documentation for how the array returned from ObservationFromGrid is orientated in terms of N/S/E/W
In part 7 of the tutorial for Malmo: https://microsoft.github.io/malmo/0.17.0/Python_Examples/Tutorial.pdf, They use the curly quotes: o agent_host.sendCommand(“hotbar.9 1”) # press the key o agent_host.sendCommand(“hotbar.9 0”) # release the key. These quotes cause a compilation errors which can be a very confusing error to figure out. Is there a way to get this fixed? Should I PR this? (All that needs to be done is exchange the curly quotes for the straight ones)
@satyamedh Im not sure how alive this chat is. I checked out your code. I think I'll be able to fix your problem, but I'd need your entire code (including xml files). As the SO people said, a minimum reproducible example is needed.
looking for some educator resource to teach AI in Minecraft
@satyamedh ill try to get to it soon, perhaps tmrw
@satyamedh yep, can you describe your current problem precisely?
Time to re-setup TF lol
Hello All! I am a beginner with the Malmo project. I am interested in using it for a user study. Does anyone know what Malmo offers in terms of data collection for multi-player? I'd like to run sessions with multiple players and collect data about people's behavior in Minecraft. I'd appreciate any help you all can give.
Hey everyone, I'm currently working on modifying the eating_1.xml mission sample. I was wondering if there is a way to make my agent keep moving. It stops now and then but that is not nice when you're trying to collect things. I'm sure it has to do with the ContinuousMovementCommands functions but I'm just not exactly sure what to do
Hi all, I'm a student who just came upon this project a few day ago. I've taken a tensorflow for beginner course and I'm wondering if Malmo can be implemented with Tensorflow, just a simple demonstration will do. Specifically, I'm trying to modify tutorial_6 to use tf_agent (https://www.tensorflow.org/agents/overview). This is the closest thing I can find online explaining how to use tensorflow for reinforcement learning. Any help or advice is appreciated.
How hard would it be to create a server version of the Malmo client mod? It would be great to have a hosted server with Malmo agents that users could interact with.
Hi, I am new to malmo platform . I want to come up with a simple house layout. Can anyone help me?
Hey, the "use 1" command in DiscreteMovementCommands is not working for me. Has anyone else faced this problem, and is there any solution to it?
I getting error while I am trying to use a different xml file in tutorial8.py?
Error starting mission: Failed to find an available client for this mission - tried all the clients in the supplied client pool.