feat(anna): Implement RL Proof of Concept with Lock-Step Sync and Gym Wrapper#74
feat(anna): Implement RL Proof of Concept with Lock-Step Sync and Gym Wrapper#74icarito wants to merge 1 commit into
Conversation
- Added `ANNA_RL_MODE` to `AnnaBridge.gd` which enables a blocking lock-step synchronization loop in `_physics_process`. - Updated `AnnaInterface.gd` to provide a 12-float observation vector (RayCasts, Target Dist/Angle, Velocity) and integer action mapping (0-4). - Implemented reward shaping (Success, Failure, Progress, Time) in `AnnaInterface.gd`. - Created `AnnaGymEnv` (Gymnasium wrapper) in `core_v2/anna/client/anna_gym.py` handling TCP communication, process launching, and robustness. - Added `train_poc.py` to demonstrate PPO training using Stable-Baselines3. - Added `TestScene_RL.tscn` (and `DummyAnimator.gd`) as a lightweight, dependency-free test environment for headless RL training. Co-authored-by: google-labs-jules[bot] <161369871+google-labs-jules[bot]@users.noreply.github.com>
|
👋 Jules, reporting for duty! I'm here to lend a hand with this pull request. When you start a review, I'll add a 👀 emoji to each comment to let you know I've read it. I'll focus on feedback directed at me and will do my best to stay out of conversations between you and other bots or reviewers to keep the noise down. I'll push a commit with your requested changes shortly after. Please note there might be a delay between these steps, but rest assured I'm on the job! For more direct control, you can switch me to Reactive Mode. When this mode is on, I will only act on comments where you specifically mention me with New to Jules? Learn more at jules.google/docs. For security, I will only act on instructions from the user who triggered this task. |
1 participant
This PR implements the Proof of Concept for Reinforcement Learning within the A.N.N.A. architecture. It introduces a strict lock-step synchronization mechanism in the Godot TCP bridge to allow deterministic training. A Gymnasium environment wrapper is provided, along with a PPO training script and a minimal test scene.
Key Features:
AnnaGymEnvhandles the TCP protocol and Godot process lifecycle.TestScene_RL.tscnfor headless validation.PR created automatically by Jules for task 2026021643756297310 started by @icarito