Benchmark and Dataset for Probabilistic Prediction of Interactive Human Behavior
Accurate prediction of probabilistic and interactive human behavior is a prerequisite to enable full autonomy of mobile robots (e.g., autonomous vehicles) in complex scenes. To enable accurate predictions, two fundamental problems should be addressed: 1) datasets of human behavior and motions in interactive tasks and scenarios, and 2) evaluation metrics and benchmarks for extensive prediction models/algorithms. Datasets are the most important asset since they provide sources for both model learning/training and validation. Similarly, evaluation metrics and benchmarks are also of fundamental importance since they provide not only criteria but also guidance for the design of prediction algorithms. Currently, the research community is still on its way to build high-quality datasets containing interactive human behavior, such as human-driven vehicles, pedestrians, cyclists, etc. Also, there is yet no widely accepted evaluation metric which can comprehensively quantify/evaluate the performance of different probabilistic prediction algorithms from perspectives of both data approximation and fatality/utility impacts on the autonomy of the mobile robots.
The workshop will be held on the Nov 4th, at IROS 2019, Macau. More details about the confernece can be found here. The tentative schedule can be found below. Details on speakers and talks will be posted soon!
Call for Papers
Submission is OPEN!
If you are working on any listed topics below or other topics that are closely related to our workshop, we would like to hear about your research!
The topics of interest include, but are not limited to,
- Human behavior data collection
- Manual annotation tools and automatic labeling algorithms for data processing
- Datasets focusing on human behaviors in interactive scenarios
- Social interaction models
- Human behavior models for prediction and generation
- Probabilistic prediction algorithms of behavior and motion
- Evaluation metrics for probabilistic prediction
- Data-efficiency, interpretability and generalizability of prediction algorithms
- Interpretable prediction models
- Survey focusing on datasets, prediction algorithms or evaluation metrics
- Risk analysis of prediction
Special topic: competition on propabilistic and interactive human behavior/motion prediction algorithms
To facilitate the development of probabilistic prediction algorithms for interactiver human behavior, we have constructed an international, adversarial and cooperative motion dataset (INTERACTION dataset), which includes extensive complex interactive human behaviors across different countries such as the US, Germany, China and some other countries. If you are a researcher on related topics, we welcome you to participate our workshop as a competitor for propabilistic and interactive human behavior/motion prediction using the INTERACTION dataset.
More details about the INTERACTION dataset can be found at: https://interaction-dataset.com. The dataset will be released at early July, and you can show your interest by submitting a request on the website.
- All contributions (regular papers or the challenge-oriented papers) should be prepared in PDF files using the IEEE template (for MS Word or LaTeX).
- Extended abstracts are up to 4 pages.
- Submission and deadline: The extended abstracts should be submitted via PaperPlaza as a "Late Breaking Abstracts". The deadline for the submission is August 19, 2019.
All submissions will be reviewed by our commitee members and a selection of our invited speakers. Selected papers are to be presented in the form of highlighted presentation as well as poster at the workshop.
For more information and enquiries, contact Wei Zhan (firstname.lastname@example.org) or Liting Sun (email@example.com).
With the theme of "Robots connecting people", IROS2019 aims to cover all related topics in intelligent robots, and human-robot interaction is one of the emphasized topics. To assure that the robots safely and efficiently interact with human, it is of significantly improrance for the robots to understand, perdict the human behaviors, and finally to learn to behave like human. To achieve such a goal, two fundamental assets are needed: a dataset covering interactive human behaviors and a comprehensive benchmark on evaluating different perdiction algorithms. To facilate the development of these two assets, we organize this workshop.
Researchers, educators and practitioners who are working on human-robot interaction, human behavior prediction, behavior modeling and analysis, representation learning, imitation learning, human-like behavior generation and social interaction can all be our andience.
- Wei Zhan, UC Berkeley;
- Liting Sun, UC Berkeley;
- Masayoshi Tomizuka, UC Berkeley;
Coming soon ...