SirStone 766b9e03ee feat(PPO_Bot): enemy-centered action space + reward shaping for 100% vs Target ...

- actions.nim: goto/aimTo coordinates now offset from enemy position
  (enemyX + tanh(raw) * scale) instead of absolute arena coords
  (sigmoid(raw) * arenaSize). Initial random policy defaults to
  approaching and aiming at enemy.
- training.nim: added dense reward shaping (distance closeness +
  gun bearing) to computeTickReward, doubled round reward scaling.
- PPO_Bot.nim: passes enemy position to mapActions, computes
  gun-to-enemy bearing for reward shaping.

Result: 100/100 win rate vs Target with frozen weights.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

2026-08-18 17:40:59 +02:00

..

tests

feat(PPO_Bot): persist Adam optimizer state and round counter across restarts (#35)

2026-08-18 15:40:54 +02:00

weights/latest

feat(PPO_Bot): command abstraction layer — goto/aimTo controllers (#24)

2026-08-17 19:20:40 +02:00

.gitignore

feat(PPO_Bot): reward + trajectory + GAE + PPO training (#16)

2026-08-16 15:27:15 +02:00

actions.nim

feat(PPO_Bot): enemy-centered action space + reward shaping for 100% vs Target

2026-08-18 17:40:59 +02:00

config.nims

fix(PPO_Bot): radar lock — use enemy_tracker width-lock, fix arctan2 arg order

2026-08-17 11:52:36 +02:00

controllers.nim

feat(PPO_Bot): command abstraction layer — goto/aimTo controllers (#24)

2026-08-17 19:20:40 +02:00

enemy_tracker.nim

fix(PPO_Bot): radar lock — use enemy_tracker width-lock, fix arctan2 arg order

2026-08-17 11:52:36 +02:00

network.nim

feat(PPO_Bot): enemy-centered action space + reward shaping for 100% vs Target

2026-08-18 17:40:59 +02:00

PPO_Bot

feat(PPO_Bot): command abstraction layer — goto/aimTo controllers (#24)

2026-08-17 19:20:40 +02:00

PPO_Bot.json

feat(PPO_Bot): reward + trajectory + GAE + PPO training (#16)

2026-08-16 15:27:15 +02:00

PPO_Bot.nim

feat(PPO_Bot): enemy-centered action space + reward shaping for 100% vs Target

2026-08-18 17:40:59 +02:00

PPO_Bot.nimble

feat(PPO_Bot): reward + trajectory + GAE + PPO training (#16)

2026-08-16 15:27:15 +02:00

PPO_Bot.sh

feat(PPO_Bot): reward + trajectory + GAE + PPO training (#16)

2026-08-16 15:27:15 +02:00

shell.nix

feat(PPO_Bot): reward + trajectory + GAE + PPO training (#16)

2026-08-16 15:27:15 +02:00

state_vector.nim

feat(PPO_Bot): command abstraction layer — goto/aimTo controllers (#24)

2026-08-17 19:20:40 +02:00

training

feat(PPO_Bot): full PPO RL implementation (#14, #15, #16, #17)

2026-08-16 16:34:37 +02:00

training.nim

feat(PPO_Bot): enemy-centered action space + reward shaping for 100% vs Target

2026-08-18 17:40:59 +02:00

weights.nim

feat(PPO_Bot): persist Adam optimizer state and round counter across restarts (#35)

2026-08-18 15:40:54 +02:00