Beyond Episodic Evaluation: Memory Architectural Bottlenecks in Sequential Embodied Question Answering
2026-07-23 • Robotics
Robotics
AI summaryⓘ
The authors studied how robots that answer questions about their surroundings work better when they remember things across multiple questions, instead of starting fresh each time. They found that just keeping track of where the robot has been isn’t enough, because it misses important details about what it saw. To fix this, the authors showed that robots need memory that links what they see to 3D maps, helping them remember both location and visual details. This approach helped robots give better answers while moving around less, and it also worked in real-world tests.
Embodied Question Answering (EQA)Sequential MemorySpatially Grounded Memory3D Metric MappingVisual-Semantic EvidenceOccupancy MapsNavigation EfficiencyRobot Memory Architecture
Authors
Zikui Cai, Kaushal Janga, Tan Dat Dao, Seungjae Lee, Shivin Dass, Mingyo Seo, Kaiyu Yue, Mintong Kang, Nandhu Pillai, Monte Hoover, Aadi Palnitkar, Ruchit Rawal, Ruijie Zheng, Bo Li, Yuke Zhu, Roberto Martín-Martín, Tom Goldstein, Furong Huang
Abstract
Embodied question answering (EQA) is traditionally evaluated under an episodic formulation, where agents solve each task independently and reset internal state between episodes. However, real-world robots operate continuously and must accumulate, retain, and selectively reuse information acquired from prior interactions. Despite this practical requirement, the architectural mechanisms needed to support sequential memory in EQA remain underexplored. In this work, we investigate how different memory architectures behave when EQA agents are evaluated sequentially, with multiple questions answered in the same scene while memory is carried forward across queries. We find that simply preserving existing memory is often insufficient. Agents that retain only traversability information, such as 2D occupancy maps, remember where the robot has explored but not the visual-semantic evidence needed for later questions. Agents trained on short-horizon episodic data face a different challenge: when exposed to continuous, multi-query histories, their inherited context suffers from severe temporal mismatch, rather than forming a reusable scene representation. To overcome this architectural bottleneck, we highlight the necessity of structured, spatially grounded memory: architectures that map persistent visual observations onto metric 3D geometry preserve visual-semantic evidence in a coherent scene representation. Extensive experiments in simulated environments reveal that this form of memory breaks the accuracy-efficiency tradeoff in sequential settings, simultaneously achieving higher answer accuracy and lower navigation costs. We further validate these findings on a real-world mobile robot, demonstrating that spatially grounded visual memory is critical for enabling continuous, intelligent operation in physical environments.