CAAI Artificial Intelligence Research

In recent years, embodied intelligence represents a significant shift in the way researchers approach artificial intelligence. By focusing on the integration of perception, action, and interaction with the environment, this field aims to create intelligent systems that can adapt and learn in complex, real-world scenarios. Emphasizing the physical body and its interaction with the environment allows researchers to explore intelligence in a more holistic and human-like manner. In the context of autonomous driving cars, for example, embodied intelligence involves developing vehicles that not only perceive their surroundings through sensors like cameras and LiDAR but also actively respond to dynamic traffic situations, pedestrians, and unexpected events. These intelligent agents need to make decisions based on their understanding of the environment, ensuring safety and efficiency in their interactions with other road users. Similarly, service robots equipped with embodied intelligence can navigate homes or public spaces, understand natural language commands, and interact with objects and humans. By combining visual perception, language understanding, and physical actions, these robots can assist with various tasks, enhancing their usability and practicality in real-world settings.

This special issue solicits research papers on the development of fundamental components in embodied intelligence, like vision, language, graphics, and robotics. Appropriate submissions include but are not limited to, the following forms or a combination thereof:

• Human-Centric 3D Scene Understanding: It involves empowering embodied intelligence to comprehend and adjust to the 3D environment where humans are present. This encompasses recognizing objects in the surroundings, estimating their poses, and understanding advanced concepts like affordance.
• Reinforcement Learning: Utilizing reinforcement learning for embodied agents to acquire human-like motor and cognitive skills. This approach enables robots to enhance their learning capabilities and adapt effectively in intricate real-world environments..
• Manipulation: Tackling tasks such as robotic grasping, transportation, and assembly. This includes improving a robot's manipulation capabilities and adaptability, as well as how to enable robots to collaborate better with humans or other robots.
• Mobility: Enabling robots and autonomous vehicles to move more effectively in complex environments. This includes issues such as robot navigation, obstacle avoidance, and path planning.
• Simulation: Testing and verifying robot performance and algorithms through simulation. This includes establishing more realistic simulation environments, as well as how to enable robots to learn and adapt better in realistic environments.
• ToM and LLM: Exploring the computational theory of mind (ToM), potentially with the assistance of large language models (LLMs), involves understanding human language, behavior, and emotions. This exploration aims to enhance robots' interactions with humans, enabling them to better comprehend and adapt to human interactions.

This special issue welcomes excellent papers on various research topics related to the above directions as well as other relevant fields for submission.

Editorial Team

Leading Guest Editor

Hao Zhao, Tsinghua University, China
Email: zhaohao@air.tsinghua.edu.cn

Guest Editors
(Alphabetical Order)

Hao Dong, Peking University, China
Hang Zhao, Tsinghua University, China
Jiangmiao Pang, Shanghai AI Laboratory, China
Li Yi, Tsinghua University, China
Qichao Zhang, CASIA, China
Siheng Chen, Shanghai Jiao Tong University, China
Yixin Zhu, Peking University, China
Yiyi Liao, Zhejiang University, China

Timeline

Submission due date: March 20, 2024

Submission

https://mc03.manuscriptcentral.com/ai-journal

Kindly choose for inclusion in “Special Issue on Embodied Intelligence” upon submission.