Yangzi Li

Boyu Mi, Mengchen Ma, Yifei Yao et al.

Real-world deployment of embodied agents requires active exploration, visual grounding, and interactive intent disambiguation. However, existing frameworks often rely on privileged simulator states or assume complete instructions, bypassing realistic deployment challenges. To bridge this gap, we present REAL, an agentic framework for open-world mobile manipulation. REAL establishes sim-to-real-consistent environment APIs without oracle perception and integrates a simulated user to enable human-in-the-loop interaction. Within this environment, we design diverse task compositions to drive data collection, supervised fine-tuning, and online reinforcement learning, systematically optimizing agent performance. To comprehensively evaluate this approach, we introduce REAL-Bench, a benchmark spanning 241 tasks across active exploration, visual distraction, articulated manipulation, and interactive disambiguation. Experimental results demonstrate that our trained agent outperforms leading commercial closed-source VLMs on interactive tasks with a 56.9% success rate. Further empirical analysis reveals that our hierarchical training pipeline successfully aligns the model's tool-use capabilities while maintaining robust open-vocabulary reasoning under extended exploration horizons. Finally, we deploy and evaluate our framework on a physical dual-arm mobile robot, where it achieves a 78.3% end-to-end success rate over 60 real-world episodes. These physical trials demonstrate robust zero-shot transferability to unseen household scenarios, validating that our sim-to-real-consistent design successfully bridges the reality gap for long-horizon mobile manipulation. Code is available at https://github.com/InternRobotics/REAL.

Zhilu Lian, Yangzi Li, Zhixiang Chen et al.

Working set size estimation (WSS) is of great significance to improve the efficiency of program executing and memory arrangement in modern operating systems. Previous work proposed several methods to estimate WSS, including self-balloning, Zballoning and so on. However, these methods which are based on virtual machine usually cause a large overhead. Thus, using those methods to estimate WSS is impractical. In this paper, we propose a novel framework to efficiently estimate WSS with eBPF (extended Berkeley Packet Filter), a cutting-edge technology which monitors and filters data by being attached to the kernel. With an eBPF program pinned into the kernel, we get the times of page fault and other information of memory allocation. Moreover, we collect WSS via vanilla tool to train a predictive model to complete estimation work with LightGBM, a useful tool which performs well on generating decision trees over continuous value. The experimental results illustrate that our framework can estimate WSS precisely with 98.5\% reduction in overhead compared to traditional methods.

Yangzi Li, Jyh-An Lee

The unauthorised use of data in the training of generative AI models presents significant legal challenges, particularly under intellectual property (IP) and privacy laws. These frameworks frequently grapple with the intricate relationship between data ownership and AI innovation, resulting in ongoing debates regarding optimal protection and enforceability. This article delves into considerable potential of unjust enrichment as an alternative legal doctrine for resolving disputes arising from such unauthorised data use. We explore how the concept of unjust enrichment captures the wrongfulness of unauthorised data use in a manner distinct from IP infringement and privacy violations. Furthermore, we analyse the extent to which gain-based restitution for unjust enrichment may prove more advantageous than existing remedies, including legal, equitable, and statutory options. We content that by shifting the emphasis from establishing wrongful conduct to recovering benefits obtained unjustly, unjust enrichment offers a pragmatic and equitable framework that reconciles the rights of data owners with the interests of AI developers.