After a grueling two-month competition, we are thrilled to announce the results of the PRS challenge! PRS challenge scenario is described as a "human-centered in-building embodied delivery service" in a commercial context. The challenge was conducted in a novel simulated environment, Polar Research Station (PRS), featuring robots, human characters (agents), numerous interactive items, and a complex architectural setting designed to mimic real-world human-robot interaction scenarios closely. This allowed participants to comprehensively and thoroughly evaluate the performance of existing technologies (LLM, LMM, Embodied AI) in such human-robot interaction contexts. Given the complexity of the challenge and the difficulty level of the delivery tasks, we invited teams to participate, ensuring that they would receive technical support throughout the challenge and would fully understand the business scenario of the challenge. As previously mentioned, the primary objective of this challenge was to swiftly identify superior solutions for the presented scenario. In this challenge, we successfully attracted 19 teams from top institutions, universities, and companies worldwide. Based on a range of criteria, including overall contribution, the degree of solution innovation, method efficiency, and more, we have selected the top three teams as follows:

PRS Challenge on CVPR 2024 Embodied AI Workshop Awards Winners

Prize/Ranking Team Affiliation Members
1 PDA Johns Hopkins University, University of Southern California Xingrui Wang, Feng Wang
2 Team S Imperial College London, Cambridge University Xiaoyao Qiu, Fanyi Wu
3 Renmin Dynamics Renmin University Hanfang Yang, Jinhua Su, Han Zhang
Additionally, to express gratitude for the active participation and extraordinary contributions of all teams during the challenge, the organizers have decided to present the Outstanding Contribution Award to the following teams (listed in no particular order):

PRS Challenge on CVPR 2024 Embodied AI Workshop Outstanding Contribution Award

None Team Affiliation Members
- BIRD Nanyang Technological University (NTU) Haoyu Liu, Zulun Zhu, Kedao He
- CASIA_TSINGHUA_IR Chinese Academy of Sciences Institute of Automation, Tsinghua University Xian Tao, Qu Zhen, Qu Shichen, Gao Wenbo, Gong Xinyi, Kai Chen, Tao Lu, Yinghao Cai
- CEBSIT-EmbodiedAI Center for Excellence in Brain Science and Intelligence Technology, Chinese Academy of Sciences Tielin Zhang, Shuncheng Jia, Liyuan Han
- BUPT Embodied AI Group Beijing University of Posts and Telecommunications Bin Fang and Yuhao Sun
- Neanderthals Luddy school at Indiana University Bloomington Yueru Yan, Xinyi Zhao
- BigLOL Horizon Robotics Dong Cao,Zihan, Li,James Ni, Fei Gao
- EDIN_EAI University of Edinburgh ; Google DeepMind Shangmin Guo, Biao Zhang
- X_EVA Bin Dai, Weiyuan Li, JingShen Gao
- DreamTeam North China University of Technology Bi Song, Yun Ding, Siyuan Chen, He Wan, Ang Bi
- Cx330 Institute of Computing, Chinese Academy of Sciences Zejun Yang, Zihan Liu, Yinghua Zhao, Jialong Lan
- HKUST_EAI The Hong Kong University of Science and Technology Jialun Cao, Jinting Wu
Finally, all data, documents, materials, and models from this challenge will be released collectively in the near future. At that time, everyone can participate in this permanently open challenge through email registration. Meanwhile, we are open to feedback from all participants and will update our settings and environment based on reasonable suggestions. We warmly invite all teams interested in our scenario to communicate with us and jointly explore the topic actively.

Centered Image