修改task1的策略形式化 - #2
Open
xuHaob1 wants to merge 37 commits into
Open
Conversation
spatial_c 中 final_chest 位于 abyss tile [6,4]，虽 is_visible=True 但引擎不渲染 abyss 上的 chest sprite，detect_chest_tile 返回 None，导致 agent 在 center/south 间无限弹跳。 修复：新增 probe_hidden_chest 方法扫描桥边 tile 并按 A 探测（引擎仅检查 Manhattan 邻接）；navigate_to_exit 增加 exit 旁像素对齐修复第一阶段 west exit 卡死。验证：original/spatial_a/b/c 各 3 seed 全部 100% 成功。
修复内容： 1. move_towards_aligned像素对齐阈值+1改为+0，添加卡住检测(mem_last_action_blocked) 2. detect_task3_room从硬编码位置改为全网格扫描NPC/chest 评测结果(num-envs=30 robustness suite)： task_1-4 original 72/72 + spatial 36/36 全通过 color变体0/12预期失败(策略基于精确RGB匹配) task_5代码未变仍为0%(agent_dead)
Bug 1: navigate_to_exit像素对齐方向检查(避免地图边界卡死) Bug 2-4: button_pressed/monster_cleared事件误标修复 Bug 5: detect_task5_room east/west墙体模式修正 Bug 6: chest_opened误标(用chest_adjacent and not monster_adjacent区分) Bug 7: attack_monster缺少blocked/player_px参数 task_5 original: avg_steps 400->1083, avg_reward -21.65->95.02, 全部里程碑1.0 task_1-4: original+spatial 100%, color 66.7%, 与修复前一致
baseline_policy.py: start房先杀chaser避免转向接触伤害、east跳过ambusher(ambush_range=2不激活)省~130步、west用detect_closest_monster_tile逐个击杀全部怪物修复多怪物质心bug，agent在step 1070以HP=1开完4宝箱触发world_completed main.tex: 更新结果表格(0.000到1.000)、Task 5状态说明(HP预算分析+三项优化关键+优化历程表格)、时间安排(2026-07-15条目)、总结与展望
…riant results - Fix detect_task5_room start room: use wall features instead of hardcoded object positions - Remove south room chest hardcoded fallback (8,5) -> NOOP wait for perception - Update main.tex: color 3/5 (dark/bright/inverted pass), spatial 0/3 (HP budget issue)
Replace stale eval_results.json (task_5 0%) with 500-episode robustness data (all 100%). Fix report num-envs/episode counts (10->100, 50->500), color variant description (1 grayscale -> 5 variants), and table values. Update PROJECT_PROGRESS numbers. Rewrite contribution table by actual commit records. Add Appendix A with Lean build and eval run screenshots. Remove temp debug files.
Per readme section 9: add milestone achievement table (task_3/4/5) and progress event_counts table (task_1-5) using eval_results.json data. Add student IDs to author field and contribution table per TA submission requirement.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
No description provided.