ICAPS 2025 Call for Demonstrations:

The System Demonstrations and Exhibits program at ICAPS 2025 provides an opportunity for planning and scheduling researchers and practitioners to demonstrate their state-of-the-art implementations in action. This event allows the community to experience some of the latest contributions to the field while broadening the reach of novel methods. Researchers from all sub-areas of AI planning and scheduling are encouraged to submit proposals to demonstrate their systems. Submissions will be evaluated based on their novelty, scientific or industrial contributions’ relevance to the conference theme and participants, and presentation. We welcome submissions that demonstrate novel techniques for automated planning and scheduling, as well as commercial, mature systems, and innovative applications recently deployed or in development that rely on automated planning and scheduling in a significant way. Tools for teaching planning or scheduling are also relevant. There will be an award for Best System Demonstration.

In the recent years, interest in planning has expanded beyond traditional ICAPS methods, with approaches such as reinforcement learning with learned representations and large language models (LLMs) gaining attention. We encourage demonstrations that explore these and other non-traditional methods, particularly those showcasing novel capabilities in planning.

To foster meaningful discussion, we ask that all submissions—including those using learning-based or hybrid techniques—address the following aspects:

  • Robustness: How does the system ensure that actions are applied correctly and that plans achieve their goals? If there are no explicit guarantees, how does the approach handle errors across different scenarios? An action is applied correctly if the state satisfies the preconditions for safe execution and reliably leads to the expected or possible results in cases of non-deterministic or probabilistic effects
  • Generalization: How adaptable are the methods to unseen situations or new domains?
  • Scalability: How well do the methods perform as the problem size increases?
  • Deployment Effort: What are the practical considerations for using the approach in this context, including symbolic modeling, hyperparameter tuning, data requirements, and computational costs? If a pre-trained model is used, what is the effort required to adapt it to a specific domain?

As a reference, AI planning methods published in the ICAPS main track are generally expected to be robust and generalizable. Their scalability is explicitly discussed. The deployment effort is usually made explicit by conveying the complexity of a symbolic model and how the planning method is orchestrated with the other system components.

While ICAPS has traditionally emphasized methods with strong guarantees in these aspects, we welcome submissions that introduce alternative strengths, such as adaptability and learning from experience. By welcoming demonstrations using alternative methods, we aim to enrich the planning community’s understanding of these challenges.

Submission Instructions

Potential exhibitors should submit their applications via Easychair.

The submission should consist of an extended abstract of no more than two pages using the latest AAAI template. This should include the title of the proposed demonstration, the list of authors and their affiliations, and an abstract of up to 250 words. It should describe the technical content of the demonstration, with the appropriate credits and references. Authors may upload supplementary material to describe any non-standard hardware or software requirements, or other requested logistical arrangements. We plan to provide for each demonstration a table, chairs, poster board, power strip, and a large display, but will try to accommodate special requests.

The extended abstract must include a link to a video file no more than 10 minutes long, hosted on a public service that allows streaming or downloading. The video of the demonstration is as important as the extended abstract for evaluating the suitability and merit of the proposed demonstration.

The extended abstracts and videos for accepted demonstrations will be available to conference participants and will be archived on the conference website.

Summary of Important Dates

  • Submission Deadline: Monday, June 9th.
  • Notification: July 21st.

System Demo Program Chairs

  • Jiaoyang Li, Carnegie Mellon University (jiaoyangli at cmu dot edu)
  • Hector Palacios, independent, (hectorpal at gmail.com)

Please direct all questions about the demonstrations program to the system demonstrations chairs.