Xuelong Li

Institute of Artificial Intelligence (TeleAI), China Telecom Abstract The high cost of collecting real-robot data has made robotic simulation a scalable platform for both evaluation and data generation. Yet, most existing benchmarks concentrate on simple manipulation tasks like pick-and-place, failing to capture the non-Markovian characteristics of real-world tasks and the complexity of articulated object interactions. To address this limitation, we present RuleSafe, a new articulated manipulation benchmark built upon a scalable, LLM-aided simulation framework. RuleSafe features safes with diverse unlocking mechanisms—such as key, password, and logic locks—that require distinct multi-stage reasoning and manipulation strategies. These LLM-generated rules yield non-Markovian, long-horizon tasks that demand temporal modeling and memory-based reasoning. We further propose VQ-Memory, a compact and structured temporal representation that leverages vector-quantized variational autoencoders (VQ-VAEs) to encode past proprioceptive states into discrete latent tokens. This representation effectively filters low-level noise while preserving high-level task-phase context, providing lightweight yet robust temporal cues that are compatible with existing Vision-Language-Action models (VLA). Extensive experiments on state-of-the-art VLA models and diffusion policies demonstrate that VQ-Memory consistently improves long-horizon planning, enhances generalization to unseen configurations, and achieves more efficient manipulation with reduced computational cost. Project page is https://vqmemory.github.io. 00footnotetext:, [