BetaMem improves long-term memory for robots in changing environments
BetaMem: Learning to Remember a Changing World for Embodied Agents
Artificial Intelligence
Summary
Robots and agents that interact with the world over long periods face problems remembering what happened correctly and completely. The authors introduce BetaMem, a multi-layer memory system that learns to store and update detailed and accurate records of experiences over time. BetaMem uses a combination of supervised learning and reinforcement learning to check for errors in memory and fix them automatically. This improves the robot's ability to recall past events and understand habits from long observation histories.
What this means in practice
- •For robot developers: Enable robots to remember past actions accurately over time in complex, changing environments to improve task execution and error correction.
- •For virtual assistant builders: Improve virtual assistants’ ability to recall user habits and long-past events for more personalized and context-aware interactions.
Authors
Mai Liu, He Sun, Siru Zhong, Kerun Chen, Hao Yan, Xiaohui Lv, Shuai Tao, Wulong Liu
Abstract
Long-horizon embodied agents face temporal confusion and incomplete or distorted records, recurring errors from fixed memory rules, and growing storage and retrieval costs. We present BetaMem, a framework for building and evolving traceable long-term memory through six linked layers: working, spatial, entity, episodic, experiential, and procedural memory. A dual-level write contract constrains operation executability and source support for content: supervised fine-tuning learns compliant memory operations, while reinforcement learning combines repeated sampling with teacher verification of the original observations to reward complete, accurate records and penalize omissions, distortion, and unsupported content. We formulate failure-driven policy recursive self-improvement through diagnosis along the causal sequence of memory operations and local rule repairs gated by independent replay. Atomic tools frame storage, updates, retrieval, and forgetting as cost-utility allocation under storage and read budgets. Using the open-source Qwen3.5-9B model for memory generation, BetaMem achieves overall scores of 48.08% on M3-Bench-robot, 92.08% on LoCoMo, 72.6% on OpenEQA EM-EQA, and 62.4% on EgoLifeQA, the highest among the listed memory methods under their respective evaluation protocols. Training the model for memory generation raises the contract pass rate from 50.9% to 86.2% while reducing output tokens per stored action by 47.3%. A separate Qwen3-VL-8B-Instruct evaluation on EMem-Bench achieves an average success rate of 38.4% and an average error recurrence rate of 24.9%. We perform best among evaluated methods using comparably sized open-source models. Category results show that we have advantages in integrating information across observations, recalling events over long histories, and inferring personal habits.