C Results in ALFRED's T ask and New T asks

This section contains supplementary material to support the main paper text. In the video, we show the process of creating and building the activity-context memory described in Sec. The video compares the baseline approaches to our method. We present the algorithm for creating and maintaining the ACO memory in Algorithm 1. Equation 5. Note that in practice, we normalize The remaining columns show similar "states" from THOR that our agents deem We show additional detection results to supplement Figure 1 (left). Last column shows failure cases.

C Results in ALFRED's T ask and New T asks

While Large Language Models (LLMs) can solve many NLP tasks in zero-shot settings, applications involving embodied agents remain problematic. In particular, complex plans that require multi-step reasoning become difficult and too costly as the context window grows. Planning requires understanding the likely effects of one's actions and identifying whether the current environment satisfies the goal state. While symbolic planners find optimal solutions quickly, they require a complete and accurate representation of the planning problem, severely limiting their use in practical scenarios. In contrast, modern LLMs cope with noisy observations and high levels of uncertainty when reasoning about a task. Our work presents LLM Dynamic Planner (LLM-DP): a neuro-symbolic framework where an LLM works hand-in-hand with a traditional planner to solve an embodied task. Given action-descriptions, LLM-DP solves Alfworld faster and more efficiently than a naive LLM ReAct baseline.