June 16, 2025
this is super cool!!!! like litterally loved it
This project is now in a shipable state which is good because I am out of time in SOM to work on it. For this devlog I added the ability for you to select which robot you want to play in the menu as well as fixing various Spartan robot bugs. Additionaly I made sure wildstang still works after all the changes I made to the arm and elevator nodes.
I've spent a lot of time making cargo/hatch scoring more reliable and less janky. I also tried many different combinations of settings and joints and models to try and the make the climb work. Unfortunetly it is just too unstable right now and I do not have the time to continue working on it and still get things like preloads and robot selection done so the game is actually playable. So, spartan's climb will unfortunetly not be avalible in the release.
I got cargo fully working with spartan robot. Now I am working on the climb.
I got the ability to intake and score cargo at a basic level. It works but there are still a lot of bugs and edge cases to work out.
I added the ability to score hatch pannels at all levels.
I added the ability for the spartan's robot to grab a hatch from the human player but can not score them yet.
I got the elevator and arm imported into godot and moving manually. I still have to add the climber and states but this is a good chunk done.
I got the drivetrain for the 971 robot imported into godot. I had to heavily simplify the robot RIO and 2 other parts so they would fit in the 256 max surface size for meshes imported from gltfs
I created a custom import script that removes a lot of the extra nodes that are created when importing a gltf and only keeps the meshes and collision shapes so I don't have to do that manually.
I got nearly all the cad done for the Spartan Robotics 971's 2019 robot. The modeling is done but materials still need to be done. I am hoping to be done with it before summer of making ends. I ended up making this cad in a different way than I made Wildstang's. For wildstang I recaded the whole thing from scratch but for this one I took spartan's assembly and removed all the bolts and other small extra pieces. Then I simplified some of the very intensive models like the omni wheels. This will result in a better looking robot but more preformance intensive.
Made hatch scoring easier and made the hatch grab arms move. This is for real the last update for wildstang
I added the ability to restart the match so you don't have to restart the game to play again. I also added more detail to the cargo ship and made it easier to score hatch pannels.
Today I cleaned up that last few things to get wildstang robot in a state that I'm comfortable working on the next bot. I also cleaned up the filesystem a lot and created the first demo build.
Godot 4.5 released today so I messed around with updating to that version. Unfortunely the godot-jolt pluging is not supported for 4.5 (maybe never supported) so I could not continue using the joltBasedNodes. I tried using the new nodes but I could not get the joints nearly as stable as I would like and would rather focus on adding new features instead of messing around with physics again. I kept a copy of the 4.5 version so I will come back to it eventualy but not a priority for now.
I updated the gui for the preloaded game pieces so it uses uses the real field to show it's changes and I feel it greatly improves the look
I got the climb working for the 111 robot. Because of differences between the game and real world physics that I have not been able to fix the climb is not exactly the same as their real life one where they were able to launch themselves onto the hab. Luckly they did plan to have a mechanism for their climb which is in their CAD but never got but on their irl robot so I implemented that so they can climb.
I got hatch pannels working with wildstang. It is very similar to how it works for 1323 but instead of launching the hatchpannel to score it pushes it out then drops it. There were a lot of issues with the hatch pannel clipping into things and causing weird physics things but I have worked out most of them.
I finished the cad for wildstang, I may go back through and add the pocketing but I want to know what the preformance is like without it first. I also got the cargo intaking -> manipulator working. It is essentially the same system as with the madtown robot but I did find a few bugs that I didn't catch before. For example the intakes never actually got disabled when I told them to. I also adjusted my elevator code so it supports a cascade elevator, which wildstang has, while it only supported continuous before. The elevator works when controled manually but I still need to add the setpoints and buttons to the code to get it to move to scoring positions.
I've decided to make 111's 2019 robot for the first real, well done robot because they can do evreything pretty good and I like the team. I got a good start on the cargo ground intake and passthrough but I still need to do the elevator, manipulator, and some side pannels
I added a 3rd person camera view and you can switch between 1st and 3rd person by pressing spacebar. I also added some missing tape lines and some bug fixes. Now I'm going to work on a new robot.
I created a menu screen that allows you to select where you want to start, what alliance color you want to be and what game pieces you want to start with. The gui stores the inputs then creates the field node and sets the apropriate variables. Finaly it switches to that node so you can play the game.
I improved the look of the scoreboard so it matches what was actually used. I made it by uing the image of the scoreboard provided by FIRST then tracing it with inkscape so I could import it into godot. I also got the timer working so there are distinct Sandstorm, teleop, and endgame states and stops counting score after 2min 30sec which is the length of at match
I spent a lot of time improving the graphics of the field. I used images from FIRST as decals so the field elements and game pieces look a lot more like how they did in real life which honestly makes it a lot more fun to play. The original images were jpeg so I masked out the blank white parts and converted them to transparent PNGs so they would be usable. I also fixed a bug with the hatch pannel clipping into and getting stuck in some field elements.
For this one I just fixed some bugs that caused the cargo to get stuck while intaking and otheres that caused the robot to flip. I also added the ability for the human player station to drop cargo.
Scoring now works! When hatches snap to their placements they call a function on the gui that causes it to tick up the score and counter on the GUI. When cargo intersect with the scoring collider for >1 second (so they have a chance to pass thorugh if there isn't a hatch pannel) it calls a similar function on the GUI. The gui will change in the future to look prettier (like many things currently) but it makes it a lot more fun to play right now.
I got madtown's latervator working. It uses the same system that I used for the elevator but it took a lot more time because Godot's joints don't expose the joints length so you have to calculate it yourself which took a while to get the calculations right. I tried to get physics based movement working for intaking hatchpannels like I have for the cargo but I couldn't get it working so I resorted to teleporting the hatch pannles around. Next up is getting the scoring system working.
Today I shifted control of the robot components from a bunch of if-else statements to a Finite State Machine to make the control flow easier. Also preformnce is currently looking pretty good with it running at 60fps even on my slow laptop.
I got madtown working with cargo intaking. I made the intake with some Area3Ds that apply an force on the cargo moving it through the robot into the intake. The cargo is then held against a wall until a button is pressed removing collsion from the wall and shooting the cargo out. The elevator is done with the joltSliderJoint's motor.
I have decided to make a copy of team 1323 Madtown's 2019 robot for the first robot of the game. I designed the robot in Onshape with videos of the robot as my refrence. The hardest part is that all coliders in godot must be concave so I've made the robot out of many parst to make each individual part convex.
I used Onshape to cad all of the field elements for the 2019 FRC game. To accurately create these elements, I used the dimensions given in the game manual and provided field CAD. I chose to recad all the elements instead of using the FIRST provided field cad because the provided CAD is full detail down to individual bolts which is just not required for a game and very bad for preformance. Making the top of the rocket was by far the hardest part to CAD because it required a lot of finiky lofts to make work.
I had to take a week off but I have completed a basic implementation of a continuous elevator and arm. The elevator is a set of slider joints dynamicaly created at runtime and uses collision between stages to pull the intermediate stages up. The arm uses a hinge joint. Both of these use the joint motor to control them with the desired velocity being the error between current and target psotion multiplied by a constant (essentialy the p part of a PID controller). I also fixed a major issue I had with the drivebase where after hitting a wall it would start shaking while driving by moving the frame collider lower down. Next up is to build the field for the robot to play on. I have decided to go with the 2019 game Deep Space because it is a fairly simple game.
I tuned the drivetrain so it feels a lot better to drive.
Trying to implement an elevator has been a struggle so I wanted to make an update for what I have tried. The central issue I have is that the elevator needs to react to gravity and collisions but still keep its position relative to the drive base. I started with having the elevator base be a static node and the moving stage be a rigid node with them bound together with a sliderJoint3d node. This caused the moving stage to fly away from the robot and spin uncontrollably. Switching to using Jolt physics solved the spinning but it was still not rigidly attached to the robot. I then tried a mix of animatable3d nodes and character3d nodes but they still lacked collision in many cases and any collision would be rigid when I want it to be springy. Sometimes the nodes just moved from where I placed them for no reason. I'm thinking that a lot more experimentation will be needed.
I implemented a basic swerve drive that can be driven with 'wsad' for translation and 'jl' for rotation. It uses a rigid node instead of a character node so that it can interact with walls properly. It is still a bit janky but it will work and give me a platform to test other features.
I created CAD models for the perimeter of the field in Onshape based on the field CAD for the 2025 season and the picture of the Andymark Field. I then imported them into Godot and placed them. This will provide a platform for testing driving.
A video game version of the 2019 FRC game Deep Space featuring robots from 111 Wildstang and 971 Spartan Robotics.
I improved the logic and gui for the solver mode to make it usable in different difficulties and can swap back to game mode.
The humans are now able to chop wood in the area that has been selected. There were some nasty bugs which made it take so long to implement but the system should be easily extendable to doing more then just chopping trees.
I implemented a scene that can create a selection area. I currently intend to use this to mark large areas for clearing but I will likely find more uses for it in the future.
I created the camera for the game. It pans with wsad or dragging with middle mouse button
I added a basic solver to the project. It's not good yet but it's a start. The main added difficulty to is is that unlike traditional minesweeper the color does not tell you the number of bombs around it for certain. For example, a blue rupe means that there is 1 or possibly 2 bombs around it.
I added a basic GUI to the gameplay mode so it is completely playable. You are able to switch between difficulty levels and keep track of how many rupees you have earned for that level. Next up is to work on the solver mode.
I now implemented logic so the board always generates with the right number of bombs/black rupes. The rupes also generate with the correct color relative to the number of bombs/black around them.
I created the tile scene which is clickable and holds the image for what the tile contains. I also created a script to generate a grid of tiles of variable size and random content. Now I need to work on connecting the content to a score and making all the images the same size.
A recreation of Skyward Sword's minesweeper clone with a solver.
I improved the GUI so that you can scale, translate, and rotate the image.
Today I spent a lot of time debuging why the png defiltering worked on 8bit images but not 16bit images. I finally managed to figure it out when I realized that I read the specification wrong. There is still a lot more to implement but this is one major hurdle out of the way.
I got tired of making a gui so instead I'm working on adding png parsing to this project. This would allow a user to draw a png instead of just an svg. Today I managed to get enough of the format to draw out a simple image. I need to continue implementing more of the format and implement an edge detection algorithm so it can actually be drawn.
I made the base for the gui to use the program. Currently many functions are not yet supported but using tkinter and custom tkinter it is able to use the file dialog open a svg file and run the code to turn it in to gcode. More work needs to be done on speeding up the code to display the gcode to make it more usable.
I have now added parsing for polygon and polyline geometric shapes with logic for giving lines width when necessary. This means that I have completed adding all geometric commands allowing this project to parse nearly every SVG. I now plan on either making a command line tool or GUI (or both) that will allow easy importing, positioning, and scaling of an SVG for printing.
SVG images have a stroke width property that makes lines thicker. I have now implemented this property so that images appear as expected
I have now implemented all of the path commands allowing for a large majority of SVGs to be parsed. Now I am moving on to the geometric commands like polygon and ellipse. I plan on parsing these by turning them into a path command then parsing that so limit code duplication. I also made the number of points that make up an arc dependent on the length of the arc so they don't lose quality
I implemented enough of the SVG format to turn this svg: https://www.svgrepo.com/svg/521276/docker-16 into usable gcode
Parse an SVG into GCODE that a 3d printer can draw with a pen attached
This was widely regarded as a great move by everyone.