Tag
This paper proposes a reasoning-guided learning framework for generating personalized, constraint-compliant travel packing checklists, combining symbolic rules, preference learning, and constrained optimization, achieving high recall and constraint satisfaction with real-world deployment in a production iOS app.
Current mainstream pure data-driven robot solutions suffer from low data efficiency and poor generalization. The newly proposed neuro-symbolic physical intelligence paradigm breaks down tasks into two steps: world modeling and planning. It requires only 1-10 demonstrations to learn new tasks, and its generalization ability far exceeds traditional end-to-end solutions, providing a more reliable path for general-purpose robots.
Anchor is a task-generation pipeline that addresses artifact drift in AI agent benchmarks by jointly producing instructions, environments, solutions, and verifiers from a single constraint optimization specification, yielding consistent and auditable evaluation tasks for enterprise workflows. The paper introduces ERP-Bench, a benchmark of 300 long-horizon tasks in a production ERP system, showing that frontier models satisfy explicit constraints in 26.1% of trials but reach optimal solutions in only 17.4%.
This paper presents a method to automatically generate local search neighborhoods from constraint specifications using symmetry properties, evaluated on six optimization problems.