Task-Driven Implicit Representations for Automated Design of LiDAR Systems

2025-05-28

AI Summary

LiDAR system design has long relied on manual trial-and-error, struggling to jointly satisfy task-specific requirements and hardware constraints. This paper introduces the first task-driven, fully automated LiDAR architecture search framework. It models system configurations as an implicit density distribution in a continuous six-dimensional space and pioneers the integration of flow-based generative modeling with expectation-maximization (EM) optimization—enabling differentiable, interpretable, and hard-constraint-compatible architecture search. The method unifies implicit density learning, parametric sensor modeling, and continuous-space representation. Evaluated on real-world 3D vision tasks—including facial scanning, robotic tracking, and object detection—the framework automatically synthesizes high-performance configurations. It achieves over an order-of-magnitude speedup in design efficiency while matching or exceeding the performance of expert-tuned solutions.

Technology Category

• Search and Optimization: Learning to Search
• Computer Vision: Learning & Optimization for CV
• Constraint Satisfaction and Optimization: Search

Application Category

• Search and Retrieval-Augmented AI: Large language models for search
• Economics, Online Markets and Human Computation: Architectures and workflows that use LLMs for crowd work
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

Imaging system design is a complex, time-consuming, and largely manual process; LiDAR design, ubiquitous in mobile devices, autonomous vehicles, and aerial imaging platforms, adds further complexity through unique spatial and temporal sampling requirements. In this work, we propose a framework for automated, task-driven LiDAR system design under arbitrary constraints. To achieve this, we represent LiDAR configurations in a continuous six-dimensional design space and learn task-specific implicit densities in this space via flow-based generative modeling. We then synthesize new LiDAR systems by modeling sensors as parametric distributions in 6D space and fitting these distributions to our learned implicit density using expectation-maximization, enabling efficient, constraint-aware LiDAR system design. We validate our method on diverse tasks in 3D vision, enabling automated LiDAR system design across real-world-inspired applications in face scanning, robotic tracking, and object detection.

Problem

• Automated LiDAR design for complex spatial-temporal requirements
• Task-driven implicit representations in 6D design space
• Constraint-aware sensor modeling for diverse 3D vision tasks

Innovation

• Continuous six-dimensional design space representation
• Flow-based generative modeling for implicit densities
• Expectation-maximization for constraint-aware LiDAR design