Efficient Model-Based Purification Against Adversarial Attacks for LiDAR Segmentation

2025-08-25

AI Summary

LiDAR semantic segmentation in autonomous driving is vulnerable to adversarial attacks, yet existing defense methods suffer from high computational overhead and lack lightweight, range-view–oriented solutions. Method: This paper proposes the first model-driven purification framework tailored for range-view representations. It uniquely models adversarial perturbations directly in the range-view domain, constructing a differentiable and interpretable optimization network. A mathematically grounded, lightweight purification module leverages geometric and structural priors to achieve efficient robustness enhancement. Contribution/Results: Unlike generative-defense approaches reliant on large foundation models, our method significantly reduces computational cost while outperforming both generative defenses and adversarial training on benchmarks such as SemanticKITTI. Real-world vehicle deployment demonstrates its stability and accuracy under practical operating conditions.

Technology Category

• Computer Vision: Adversarial Attacks & Robustness
• Machine Learning: Adversarial Learning & Robustness
• Search and Optimization: Adversarial Search

Application Category

• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Large language models for search

Abstract

LiDAR-based segmentation is essential for reliable perception in autonomous vehicles, yet modern segmentation networks are highly susceptible to adversarial attacks that can compromise safety. Most existing defenses are designed for networks operating directly on raw 3D point clouds and rely on large, computationally intensive generative models. However, many state-of-the-art LiDAR segmentation pipelines operate on more efficient 2D range view representations. Despite their widespread adoption, dedicated lightweight adversarial defenses for this domain remain largely unexplored. We introduce an efficient model-based purification framework tailored for adversarial defense in 2D range-view LiDAR segmentation. We propose a direct attack formulation in the range-view domain and develop an explainable purification network based on a mathematical justified optimization problem, achieving strong adversarial resilience with minimal computational overhead. Our method achieves competitive performance on open benchmarks, consistently outperforming generative and adversarial training baselines. More importantly, real-world deployment on a demo vehicle demonstrates the framework's ability to deliver accurate operation in practical autonomous driving scenarios.

Problem

• Defending 2D range-view LiDAR segmentation against adversarial attacks
• Providing lightweight purification with minimal computational overhead
• Ensuring reliable perception for autonomous driving safety

Innovation

• Efficient purification framework for 2D range-view
• Explainable network based on mathematical optimization
• Minimal computational overhead with strong resilience