VLA2026/9/14arXiv:2609.15475

P-POSEMEM: ポーズグラフ再構築下で一貫した言語接地を実現する射影的意味記憶

P-POSEMEM: Projective Semantic Memory for Consistent Language Grounding under Pose-Graph Rewrites

SLAMのポーズグラフが最適化・ループ閉じ込め・圧縮されても、言語指示が同じ物体を指し続けるよう、観測を誕生キーフレームの不変イベントとして保持しベイズ木の消去条件付きを統合する意味記憶手法を提案。

詳しい要約

1. どんなもの？

言語指示に従うロボットのsemantic memoryを、SLAMのpose graphが最適化・loop closure・圧縮で書き換わっても一貫させる手法P-POSEMEMの提案。各観測をbirth keyframeの不変イベントとして保存し、marginalizeされたkeyframeのBayes-tree elimination conditionalを保持し、pose・anchor・identityの再構成されたjoint posterior上でsemantic likelihoodを積分する。Dprojはinference-equivalentなfull graphとmarginalized graphのlanguage-goal分布間のtotal-variation defect。

2. 先行研究と比べてどこがすごい？

各detectionをworld coordinateに固定する従来のmapは、closureでanchorが動くかsolverがmarginalizeするとqueryが別物体を選ぶ。P-POSEMEMはfull-graph oracleを再現(Dproj=0)し、全てのmemory-reducing baselineに対しgoal flipsを削減。761 closuresで最大47 m書き換わる8-run campaignでDproj<10^-13、goal flips 0/288。ablationや挿入時固定coordinateはflipする。live bounded solver下でも23/288対53。

3. 技術・手法の肝は？

- 各observationをbirth keyframeのimmutable eventとして保存 - marginalizeされた各keyframeのBayes-tree elimination conditionalを保持 - pose・anchor・identityの再構成joint posterior上でsemantic likelihoodを積分 - Dproj: inference-equivalentなfull/marginalized graphのlanguage-goal分布間のtotal-variation defectで直接測定 - 共有frozen detectorでretrievalを固定し、gainをmemory consistencyに隔離

4. どうやって有効だと検証した？

- 40 HM3DSem scenes、112,000 queriesで評価 - full-graph oracleを再現(Dproj=0)、全memory-reducing baselineに対しgoal flips削減 - 8-run campaign、761 closures、最大47 m書き換えでDproj<10^-13、0/288 goal flips(eliminationがclosuresに追随) - live bounded solver下で23/288(対frozen coordinate 53) - pre-registered negative controlをDprojが検出、calibration errorとnavigation successは不変

5. 議論はある？

- Dprojとcalibration error・navigation successは異なるfailure modeを捉えることをnegative controlが示唆 - 有効性はeliminationがclosuresに追随する条件に依存(条件が崩れるとflip増) - 要旨からは計算コストやスケーラビリティの議論は不明

6. 次に読むべき論文は？

- Bayes-tree / factor graph marginalization (Kaess et al. の Bayes tree) - SLAM pose graph optimizationとloop closure (g2o, iSAM2) - HM3DSem dataset - semantic SLAM / language grounding (参照研究は要旨に明示なし)

※ AIが要旨から生成した要約です。正確性は原文をご確認ください。

著者: Ha Sier, Ali Salmasi, Mengya Xu, Haizhou Zhang, Jie Lu, Zhuo Zou, Xianjia Yu, Tomi Westerlund

分類: cs.RO

原文アブストラクト

A robot following language instructions needs its semantic memory to keep naming the same physical object while the SLAM pose graph underneath is optimized, loop-closed and compressed. Maps committing each detection to a world coordinate cannot: a closure moves the anchor it was measured from, or the solver marginalizes that anchor, and the query then selects a different object although both graphs represent the same posterior. P-POSEMEM stores each observation as an immutable event at its birth keyframe, retains the Bayes-tree elimination conditional of every marginalized keyframe, and integrates the semantic likelihood over the reconstructed joint posterior of poses, anchors and identities. Dproj, the total-variation defect between the language-goal distributions of inference-equivalent full and marginalized graphs, measures this directly. Over 40 HM3DSem scenes and 112,000 queries, P-POSEMEM reproduces the full-graph oracle (Dproj = 0) and reduces goal flips against every memory-reducing baseline. On an eight-run campaign whose 761 closures rewrote the map by up to 47 m, Dproj stays below 10^-13 with 0/288 goal flips when elimination follows the closures, where every ablation and a coordinate committed at insertion flip goals it does not; under a live bounded solver the same memory flips 23/288 against 53 for that frozen coordinate. A pre-registered negative control is detected by Dproj while leaving calibration error and navigation success unchanged, indicating that these measures capture distinct failure modes. Retrieval is held fixed by a shared frozen detector, isolating the gain to memory consistency. Code and data: https://anonymous.4open.science/r/posemem-2328/.

arXivで原文を読む →arXiv概要ページ