LensDesigner: A Self-Improving Agent for Optical Lens Design

2026-09-24

AI Summary

This study addresses the reliance on manual expertise in optical lens design and the difficulty of automated methods in traversing non-convex parameter spaces by proposing an autonomous agent framework. Emulating expert workflows, this framework integrates interactive physical simulation to achieve automated design and iterative optimization. It introduces a novel self-evolution mechanism and curriculum-guided strategy, leveraging large-scale lens libraries to overcome cold-start challenges while autonomously accumulating and reusing heuristic knowledge. Furthermore, it incorporates optics-aware retrieval and macro-level orchestration algorithms to enhance search efficiency. Evaluated on the LensArena benchmark, the proposed method significantly outperforms existing baselines, substantially improving both design success rates and optimization efficiency.

Abstract

Optical lens design is a complex, non-convex optimization challenge that relies heavily on human experience and intuition. Existing optimized-based automatic lens design methods struggle to navigate this vast parameter space without meticulous manual tuning. In this paper, we present LensDesigner, an autonomous agent framework that mirrors the problem-solving workflow of expert opticians. To overcome the initial cold start problem, we construct LensLib100K, an extensive optical lens library, and employ Optics-Aware Retrieval to supply physically valid structural seeds. Within an interactive physical simulation environment, the agent executes macroscopic orchestration while receiving immediate optical feedback. Furthermore, we introduce a continuous self-evolving mechanism guided by a curriculum agent. By iteratively solving design tasks with progressively increasing difficulty, the agent autonomously extracts, accumulates, and reuses design heuristics, effectively evolving its optical lens design expertise over time. At the evaluation level, we introduce LensArena, a standardized evaluation benchmark comprising $120$ diverse optical design tasks, covering extreme configurations. Extensive experiments on this benchmark demonstrate that LensDesigner significantly outperforms publicly available baseline algorithms, achieving superior success rates and optimization efficiency. We hope this work sheds light on the emerging field of intelligent optics. The code will be publicly available.

Problem

• optical lens design
• non-convex optimization
• automatic lens design
• parameter space exploration

Innovation

• Autonomous Agent
• Self-Evolving Mechanism
• Optics-Aware Retrieval
• Curriculum Learning
• Optical Lens Design