Call for Papers
We invite original contributions on emerging challenges including, but not limited to, the following tracks.
Large-scale embodied perception and world models
- Urban-scale foundation models: efficient deployment of VLA models for long-horizon navigation in non-stationary city environments.
- Physical intelligence and multi-modal world models for anticipating dynamic traffic and pedestrian patterns.
- Distributed spatial AI and SLAM: cooperative mapping and localization across heterogeneous agents via mobile ad-hoc networks.
- Active sensing and perception: information-theoretic viewpoint/sensor selection to reduce uncertainty in smart cities.
Networked embodied action and coordination
- Communication-aware reinforcement learning under time-varying latency and packet loss in MANETs.
- Collaborative manipulation and transport: multi-robot coordination for complex urban tasks.
- Cross-platform sim-to-real transfer from urban digital twins to physical ad-hoc robotic fleets.
- Semantic inter-agent communication: exchanging intent and affordances instead of raw sensor data.
Edge-native systems for embodied intelligence
- On-device inference for VLA models: hardware-aware optimization (quantization, pruning) for edge NPUs and mobile SoCs.
- Split-computing for embodied loops: dynamic partitioning between mobile agents and smart city edge servers.
- Energy-efficient embodied autonomy: balancing high-level reasoning with physical energy constraints.
- Real-time reasoning at the extreme edge for micro-embodied agents.
Secure and trustworthy embodied systems (special focus)
- Physical-world adversarial robustness (e.g., patch attacks on traffic signs, adversarial urban textures).
- Safe-critical control in crowded spaces: verifiable RL and safety layers for human–robot interaction.
- Privacy-preserving embodied analytics: federated learning without compromising citizen privacy.
- Resilient embodied mesh: safety and mission integrity under partitions or attacks on ad-hoc routing.