Publication Type
Conference Proceeding Article
Version
publishedVersion
Publication Date
5-2025
Abstract
In real-world sequential decision making tasks like autonomousdriving, robotics, and healthcare, learning from observed state-action trajectories is critical for tasks like imitation, classification,and clustering. For example, self-driving cars must replicate humandriving behaviors, while robots and healthcare systems benefitfrom modeling decision sequences, whether or not they come fromexpert data. Existing trajectory encoding methods often focus onspecific tasks or rely on reward signals, limiting their ability togeneralize across domains and tasks.Inspired by the success of embedding models like CLIP andBERT in static domains, we propose a novel method for embeddingstate-action trajectories into a latent space that captures the skillsand competencies in the dynamic underlying decision-making pro-cesses. This method operates without the need for reward labels,enabling better generalization across diverse domains and tasks.Our contributions are threefold: (1) We introduce a trajectory em-bedding approach that captures multiple abilities from state-actiondata. (2) The learned embeddings exhibit strong representationalpower across downstream tasks, including imitation, classification,clustering, and regression. (3) The embeddings demonstrate uniqueproperties, such as controlling agent behaviors in IQ-Learn and anadditive structure in the latent space. Experimental results confirmthat our method outperforms traditional approaches, offering moreflexible and powerful trajectory representations for various applica-tions. Our code is available at https://github.com/Erasmo1015/vte.
Keywords
Representation Learning, Sequential Decision Making
Discipline
Artificial Intelligence and Robotics
Research Areas
Intelligent Systems and Optimization
Areas of Excellence
Digital transformation
Publication
AAMAS '25: Proceedings of the 24th International Conference on Autonomous Agents and Multiagent Systems, Detroit, Michigan, USA, May 19-23
First Page
858
Last Page
866
Identifier
10.5555/3709347.3743604
Publisher
ACM
City or Country
New York
Citation
GE, Zichang; CHEN, Changyu; SINHA, Arunesh; and VARAKANTHAM, Pradeep.
On learning informative trajectory embeddings for imitation, classification and regression. (2025). AAMAS '25: Proceedings of the 24th International Conference on Autonomous Agents and Multiagent Systems, Detroit, Michigan, USA, May 19-23. 858-866.
Available at: https://ink.library.smu.edu.sg/sis_research/10786
Creative Commons License
This work is licensed under a Creative Commons Attribution-NonCommercial-No Derivative Works 4.0 International License.
Additional URL
https://doi.org/10.5555/3709347.3743604