Submitted successfully

Special Session

 

 

  Special Session Ⅰ: Embodied Intelligence-Driven Marine Remote Sensing and Underwater Visual Perception

 

Session Chairs:  Jieyu Yuan, Nankai University, China

                                   Chongyi Li, Nankai University, China

 

Keywords: Underwater Vision, Underwater Vision, Multimodal Perception, 3D Reconstruction, Visual Navigation, Dataset and Benchmarking

 

Information: 

This special topic focuses on embodied intelligence-driven perception and modeling for marine remote sensing and underwater visual observation. It addresses key challenges in complex marine environments, including imaging degradation, multi-scale target perception, scene understanding, and 3D environmental modeling. The scope covers marine and underwater imaging modeling, multimodal data fusion, object detection and semantic understanding, 3D reconstruction and real-time rendering, underwater visual navigation and localization, as well as dataset construction, simulation systems, and benchmarking. The goal is to promote the development and real-world deployment of AI-driven marine and underwater visual perception technologies.

 

Topics of interest include but are not limited to:

 

  • Degradation Modeling, Denoising, Enhancement, and Optical Correction for Marine Remote Sensing and Underwater Imaging

  • Marine Imaging Technologies and Multimodal Perception and Fusion Methods

  • Marine and Underwater Object Detection, Recognition, and Classification

  • Underwater Scene Understanding and Semantic/Instance Segmentation

  • 3D Reconstruction, Modeling, and Real-Time Rendering of Marine Environments

  • Underwater Visual Navigation, Localization, and Environmental Mapping

  • Lightweight and Real-Time Algorithms for Marine Visual Perception

  • Large Language Models and Foundation Models for Marine Visual Perception, Understanding, and Cross-Modal Analysis

  • Marine and Underwater Vision Datasets, Simulation Systems, and Benchmarking

  • Quality Assessment and Evaluation of Marine Environment Understanding and Underwater Visual Perception

 

  Special Session Ⅱ: Brain-Inspired Robot Visual Detection and Spatial Perception

 

Session Chairs:  Mingliang Zhou, Chongqing University, China

                                   Xuekai Wei, Chongqing University, China

 

Keywords: Robot Vision, Object Detection, Spatial Perception, Brain-Inspired Computing, Visual Processing, Spatial Memory

 

Information: 

This session focuses on leveraging insights from neural mechanisms of the human brain to overcome key challenges in robot visual detection and spatial perception. It explores how brain-inspired principles such as visual processing pathways, spatial memory formation, and attentional modulation can be incorporated into the design of robotic visual detection and spatial perception models. The session emphasizes the potential of brain-inspired approaches to improve object detection accuracy, 3D spatial understanding in complex environments, robust navigation, and dexterous manipulation, thereby enabling a shift from passive sensing to active perception and intelligent decision-making, and fostering more natural and efficient human-robot collaboration.

 

Topics of interest include but are not limited to:

 

  • Brain-Inspired Object Detection Methods for Robots

  • Bio-Inspired Robot Perception and Environmental Understanding

  • Robot Spatial Reasoning and Autonomous Navigation

  • Scene Understanding by Integrating Visual and Spatial Relations

  • Active Visual Detection and Perception

  • Neuromorphic Vision Sensors Applications

  • Spiking Neural Networks for Visual Reasoning

  • Object Detection and Planning in Dynamic Environments

 

  Special Session Ⅲ: Unsupervised Learning in the Era of Foundation Models: For Robotics and Beyond

 

Session Chairs:  Wenhui Wu, Chongqing University, China

                                    Yuheng Jia, Southeast University, China

 

Keywords: Foundation Model, Unsupervised Learning, Self-Supervised Learning, Representation Learningg, Clustering

 

Information: 

In recent years, the emergence of foundation models (FMs) has provided powerful new tools for unsupervised problems, reshaping our research paradigm. This session aims to comprehensively cover the field of unsupervised learning for robotics and beyond, spanning from the development of novel classical algorithms to the transformative impact of modern foundation models. We welcome submissions that advance both the theory and practice of traditional unsupervised learning methods, as well as those investigating how large-scale pre-training can lead to more robust and generalizable representations. We will pay particular attention to how these representations can revolutionize classical tasks such as clustering (e.g., deep clustering, scalable clustering), while also seeking new paradigms for unsupervised tasks such as anomaly detection and density estimation. Topics of interest include, but are not limited to: the theoretical foundations of unsupervised learning, novel clustering algorithms, unsupervised representation learning, self-supervised methods for pre-training, and research on applying foundation models to enhance unsupervised tasks (particularly in scenarios such as robot perception, learning, and control). This session aims to provide a platform for disseminating cutting-edge research that pushes the boundaries of various unsupervised learning developments.

 

Topics of interest include but are not limited to:

 

  • Theoretical Foundations of Unsupervised Learning

  • Novel Clustering Algorithms and Methods

  • Innovations in Classical Unsupervised Tasks

  • Unsupervised Representation Learning

  • Research on Self-Supervised Learning Algorithms

  • Unsupervised Methods for Foundation Models

  • Applied Research in Unsupervised Learning

  • Unsupervised Learning for Complex Data

  • Trustworthy Unsupervised Learning

  • Unsupervised Learning for Robotics

 

Special Session Ⅳ: Reinforcement Learning for Control and Decision-Making in Autonomous Robots

 

Session Chairs:  Shan Xue, Hainan University, China

                                    Lu Dong, Southeast University, China

 

Keywords: Reinforcement Learning, Control and Decision-Making, Autonomous Robots, Adaptive Dynamic Programming, Multi-Agent Systems

 

Information: 

This special session, "Reinforcement Learning for Control and Decision-Making in Autonomous Robots," focuses on theoretical methods and practical applications of reinforcement learning in autonomous robotic systems. As robots operate in increasingly complex and dynamic environments, achieving efficient and robust control and decision-making becomes a central challenge. Reinforcement learning, as an interaction-based learning framework, provides essential tools for autonomous control and policy optimization, while multi-agent collaboration, distributed learning, and data-driven approaches offer solutions for decision-making in complex scenarios.
The session aims to bring together the latest advances in reinforcement learning, control theory, and decision-making, covering robotic control, planning, cooperation, multi-agent systems, and real-time decision-making in dynamic and complex environments. It provides a platform for academic and industrial exchange to promote intelligent control and decision-making technologies for autonomous robots.

 

Topics of interest include but are not limited to:

 

  • Reinforcement Learning for Control and Decision-Making

  • Adaptive Dynamic Programming

  • 3Multi-Agent Reinforcement Learning

  • Robotic Control and Planning

  • Safe Reinforcement Learning and Robust Control

  • Data-Driven Reinforcement Learning

 

  Special Session Ⅴ: Environmental Perception, Navigation, and Intelligent Operation for Underwater Robotics

 

Session Chairs:  Rongxin Zhang, Xiamen University, China

                                   Tiesong Zhao, Fuzhou University, China

                                    Ying Fang, Fuzhou University, China

 

Keywords: Underwater Robotics, Underwater Vision, Sonar Image Processing, Underwater Image Enhancement, Underwater Object Detection

 

Information: 

Underwater robotics is a vital branch of service robots, playing irreplaceable roles in ocean exploration, environmental monitoring, infrastructure maintenance, and rescue operations. However, the complex underwater environment, characterized by optical scattering, absorption, and acoustic propagation challenges, severely impedes the "eyes" of robots—their visual and sonar perception systems. This special session aims to gather the latest research advances in this field, focusing on environmental perception and understanding technologies crucial for underwater robots. Key topics include enhancement, restoration, quality assessment, object detection, and scene understanding for underwater optical and sonar imagery, extending to robust perception-based underwater SLAM, path planning, and intelligent manipulation. We invite researchers to contribute and discuss how to advance the perceptual and operational capabilities of underwater robots in complex real-world scenarios.

 

Topics of interest include but are not limited to:

 

  • Underwater Optical Image Enhancement and Restoration

  • Underwater Acoustic (Sonar) Image Processing and Interpretation

  • Underwater Image Quality Assessment and Benchmarks

  • Underwater Object Detection, Recognition, and Tracking

  • Underwater SLAM (Simultaneous Localization and Mapping)

 

More conference sessions to be updated soon.