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

docs

docs(research): goto controller algorithm for issue #20

2026-08-17 17:04:57 +02:00

GotoTest

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

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

PPO_Bot

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

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

research

research: RL algorithm choice — recommend PPO for Tank Royale bot

2026-08-15 22:07:47 +02:00

spike/arraymancer-test

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

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

tools/battle_runner

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

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

CLAUDE.md

chore: add agent skills configuration

2026-08-15 19:26:43 +02:00

S

Description

No description provided

104 MiB

Languages

Nim 73.7%

Python 18%

Shell 3.7%

Java 3.5%

HTML 1%

Other 0.1%