{"id":1886,"date":"2025-11-16T10:29:58","date_gmt":"2025-11-16T10:29:58","guid":{"rendered":"https:\/\/scipapermill.com\/index.php\/2025\/11\/16\/autonomous-drivings-next-gear-unifying-perception-planning-and-safety-with-ai-2\/"},"modified":"2025-12-28T21:20:41","modified_gmt":"2025-12-28T21:20:41","slug":"autonomous-drivings-next-gear-unifying-perception-planning-and-safety-with-ai-2","status":"publish","type":"post","link":"https:\/\/scipapermill.com\/index.php\/2025\/11\/16\/autonomous-drivings-next-gear-unifying-perception-planning-and-safety-with-ai-2\/","title":{"rendered":"Autonomous Driving’s Next Gear: Unifying Perception, Planning, and Safety with AI"},"content":{"rendered":"

Latest 50 papers on autonomous driving: Nov. 16, 2025<\/h3>\n

Autonomous driving (AD) systems are rapidly evolving, driven by groundbreaking advancements in AI and Machine Learning. From robust perception in challenging conditions to intelligent, safe decision-making and efficient simulation, the research landscape is buzzing with innovation. This post dives into recent breakthroughs, synthesizing key insights from a collection of papers that push the boundaries of what\u2019s possible in AD.<\/p>\n

The Big Idea(s) & Core Innovations<\/h3>\n

One central theme is enhancing robust perception<\/strong> under various real-world challenges. For instance, \u201cDGFusion: Dual-guided Fusion for Robust Multi-Modal 3D Object Detection<\/a>\u201d from Affiliation 1, Affiliation 2, and Affiliation 3, introduces a dual-guided fusion approach to improve 3D object detection by leveraging cross-modal interactions between LiDAR and camera data, especially in adverse conditions. Similarly, \u201cMonoCLUE : Object-Aware Clustering Enhances Monocular 3D Object Detection<\/a>\u201d by researchers from Yonsei University, addresses object detection in occluded scenes using local clustering and generalized scene memory. Complementing this, \u201cHD2<\/sup><\/span>-SSC: High-Dimension High-Density Semantic Scene Completion for Autonomous Driving<\/a>\u201d from Peking University tackles the input-output dimension and annotation-reality density gaps in camera-based semantic scene completion, significantly improving 3D scene understanding.<\/p>\n

Beyond basic perception, safety and resilience<\/strong> are paramount. The \u201cArgus: Resilience-Oriented Safety Assurance Framework for End-to-End ADSs<\/a>\u201d by the Argus Research Group proposes runtime safety checks to detect and mitigate risks in real-time. This is crucial for avoiding issues highlighted in \u201cInvisible Triggers, Visible Threats! Road-Style Adversarial Creation Attack for Visual 3D Detection in Autonomous Driving<\/a>\u201d from Xi\u2019an Jiaotong University, which exposes vulnerabilities to stealthy adversarial attacks that induce false positives. The need for robust accident anticipation is addressed by \u201cPredict and Resist: Long-Term Accident Anticipation under Sensor Noise<\/a>\u201d by researchers from the University of Macau and Zhejiang University, which uses diffusion-based denoising with actor-critic models for earlier, more reliable predictions under sensor noise.<\/p>\n

Intelligent planning and decision-making<\/strong> are also seeing rapid progress, often driven by Large Language Models (LLMs). \u201cLLM4AD: Large Language Models for Autonomous Driving \u2013 Concept, Review, Benchmark, Experiments, and Future Trends<\/a>\u201d by a team from Harbin Institute of Technology, offers a comprehensive review of LLMs as decision-making \u2018brains\u2019. \u201cFLAD: Federated Learning for LLM-based Autonomous Driving in Vehicle-Edge-Cloud Networks<\/a>\u201d by researchers from Northeastern University, China and University of Victoria, Canada, introduces a privacy-preserving federated learning framework that leverages LLMs across vehicle-edge-cloud networks for efficient and adaptive driving. Further refining LLM integration, \u201cAdaDrive: Self-Adaptive Slow-Fast System for Language-Grounded Autonomous Driving<\/a>\u201d by The Chinese University of Hong Kong, Shenzhen and Baidu Inc., dynamically determines when and how LLMs contribute to decision-making, balancing high-level reasoning with real-time efficiency.<\/p>\n

Lastly, advancements in simulation and data generation<\/strong> are critical for testing and training. \u201cReGen: Generative Robot Simulation via Inverse Design<\/a>\u201d from Massachusetts Institute of Technology and Toyota Research Institute, uses LLMs to synthesize diverse and controllable scenarios for robot simulations. \u201cD-AWSIM: Distributed Autonomous Driving Simulator for Dynamic Map Generation Framework<\/a>\u201d by M. Tsukada et al., proposes a distributed simulator for dynamic map generation, enhancing accuracy and adaptability in complex environments.<\/p>\n

Under the Hood: Models, Datasets, & Benchmarks<\/h3>\n

Recent research heavily relies on and contributes to critical resources for advancing autonomous driving:<\/p>\n