Abstract of the track

The Internet of Things and intelligent edge computing are becoming essential components of modern digital, industrial, and cyber-physical systems. The growing number of cameras, microphones, industrial sensors, robots, drones, and autonomous devices creates rising demand for fast, reliable, secure, and energy-efficient data processing. Traditional cloud-centred architectures are not always suitable for applications requiring immediate response, continuous operation, privacy, or resilience to limited connectivity. Intelligent edge computing addresses these challenges by moving artificial intelligence and data processing closer to the source of information.

Background

This track focuses on integrating heterogeneous modalities such as vision, audio, vibration, radar, LiDAR, ultrasound, inertial, and environmental data to make edge systems more accurate, context-aware, and robust than single-sensor solutions. Research hotspots include lightweight multimodal models for microcontrollers, embedded GPUs, FPGAs, NPUs, and dedicated AI accelerators; model quantization, pruning, and knowledge distillation; TinyML, federated learning, neuromorphic and event-based computing, and spiking neural networks; sensor fusion, real-time analytics, and energy-efficient edge platforms.

Subject & Research Domains

Subject: Internet of Things Systems, Multimodal Edge AI, and Intelligent Edge Computing

Integrating heterogeneous modalities such as vision, audio, vibration, radar, LiDAR, ultrasound, inertial, and environmental data to make edge systems more accurate, context-aware, and robust than single-sensor solutions.

Research Hotspots:

• Lightweight multimodal models for microcontrollers, embedded GPUs, FPGAs, NPUs, and dedicated AI accelerators
• Model quantization, pruning, and knowledge distillation
• TinyML, federated learning, neuromorphic and event-based computing, and spiking neural networks
• Sensor fusion, real-time analytics, and energy-efficient edge platforms

Topics

Topics include, but are not limited to:

• Multimodal Internet of Things systems and sensor fusion on edge devices
• Intelligent edge computing architectures and edge-fog-cloud collaboration
• TinyML and machine learning on microcontrollers
• Lightweight, energy-efficient, and hardware-aware neural networks
• Real-time computer vision, object, event, and anomaly detection
• Audio-visual learning and multimodal perception
• Integration of visual, acoustic, vibration, inertial, environmental, and location data
• Edge AI for robotics, drones, autonomous vehicles, and mobile systems
• Structural health monitoring and critical infrastructure protection
• Predictive maintenance, industrial diagnostics, and smart manufacturing (IIoT)
• Smart cities, intelligent transportation, energy systems, and smart grids
• Edge AI for healthcare, remote diagnostics, and security/defence applications
• Federated, distributed, and collaborative learning
• Neuromorphic computing, event-based vision, and spiking neural networks
• FPGA, GPU, NPU, and ASIC accelerators for Edge AI
• Quantization, pruning, compression, and hardware-aware optimization
• Explainable, secure, and trustworthy multimodal Edge AI
• Experimental prototypes, datasets, benchmarking frameworks, and industrial case studies

Recommended Invited Speaker

To be confirmed