Ad: Get your own Makera Canvera Air Desktop CNC Machine today
Makera Official INS
Join in Makera Facebook Group - Carvera Air
Makera Official YouTube
#Makera #BestDesktopCNC #CarveraAir #CNCMachine
In this video I’m going to train AI to learn how to make BattleBot Combat Robots fight autonomously and then build them in real life.
Yes they’re made out of vacuum cleaners – called Henry and Harry – because they already have faces which makes them the ideal platform for embodied AI.
I'm using Pygame, Gymnasium, Stable_Baselines3 and Pytorch to train the model.
The first thing we need to do though, is train an artificial neural network to learn how to make BattleBots fight each other autonomously.
Gymnasium doesn’t have any graphical environment, so we’re going to use the game I already wrote for that.
Gymnasium itself doesn’t have any graphical environment, so I’m going to use the game I wrote to take care of that.
Gymnasium is actually only the game wrapper, allowing the game to be reset for each match, and step through the game 20 times a second returning new observations in each step.
Those observations are the robots’ own velocity, health, weapon extension, where the centre of the play area, and the opponent are from the robot’s point of view, the opponent’s relative heading.
To actually build the Neural network I’m using Stable-Baselines3 which is a learning algorithm, and that uses Pytorch underneath because the Neural net it builds is a Pytorch model and Pytorch does the maths for the leaning updates.
Makera Black Friday — Lowest Prices of the Year!
Carvera: $4,799 (Regularly $5,799 — Save $1,000)
Carvera Air: $1,999 (Regularly $2,699 — Save $700)
Z1: $1,249 (Regularly $1,399 — Save $150)
Z1 Pro: $1,649 (Regularly $1,799 — Save $150)
🎁 Extra Perks from October 27 to November 4
Free Side Clamp Set ($59.99 value)
Free-order giveaway
Early shipping
Makera Official INS
Join in Makera Facebook Group - Carvera Air
Makera Official YouTube
#Makera #BestDesktopCNC #CarveraAir #CNCMachine
In this video I’m going to train AI to learn how to make BattleBot Combat Robots fight autonomously and then build them in real life.
Yes they’re made out of vacuum cleaners – called Henry and Harry – because they already have faces which makes them the ideal platform for embodied AI.
I'm using Pygame, Gymnasium, Stable_Baselines3 and Pytorch to train the model.
The first thing we need to do though, is train an artificial neural network to learn how to make BattleBots fight each other autonomously.
Gymnasium doesn’t have any graphical environment, so we’re going to use the game I already wrote for that.
Gymnasium itself doesn’t have any graphical environment, so I’m going to use the game I wrote to take care of that.
Gymnasium is actually only the game wrapper, allowing the game to be reset for each match, and step through the game 20 times a second returning new observations in each step.
Those observations are the robots’ own velocity, health, weapon extension, where the centre of the play area, and the opponent are from the robot’s point of view, the opponent’s relative heading.
To actually build the Neural network I’m using Stable-Baselines3 which is a learning algorithm, and that uses Pytorch underneath because the Neural net it builds is a Pytorch model and Pytorch does the maths for the leaning updates.
Makera Black Friday — Lowest Prices of the Year!
Carvera: $4,799 (Regularly $5,799 — Save $1,000)
Carvera Air: $1,999 (Regularly $2,699 — Save $700)
Z1: $1,249 (Regularly $1,399 — Save $150)
Z1 Pro: $1,649 (Regularly $1,799 — Save $150)
🎁 Extra Perks from October 27 to November 4
Free Side Clamp Set ($59.99 value)
Free-order giveaway
Early shipping
This website uses cookies
Comments 0
Be the first to comment!