Control Architectures Theory (CAT) Workshop - CDC 2026

Organizers

Relevant Info and Links

Motivation and Objectives

The design and control of complex systems is one of the defining engineering challenges of this century. These systems are complex not only because of the intricacy of their individual components, but, more importantly, because of the intricate interactions among components, spanning multiple physical domains, temporal and spatial scales, and layers of abstraction. To illustrate the kinds of systems we are interested in, consider a sensor embedded in an autonomous vehicle, governed by a finely tuned circuit; the vehicle itself, which integrates a layered autonomy stack composed of hardware, perception, planning, and control modules; a fleet of such vehicles operating in a shared environment, coordinated to meet high-level objectives; and, finally, a broader urban mobility system in which this fleet interacts with public transit and human behavior. Each level is already complex in isolation, yet the most pressing design and control challenges arise from their interdependencies.

This workshop is motivated by the need for a unified theory of layered control architectures (LCAs) applicable across domains such as power systems, communication networks, autonomous robotics, biological collectives, and human sensorimotor control. Despite the diversity of these systems, they exhibit shared structural characteristics and recurrent control challenges, yet the tools to address them remain fragmented and domain-specific. While foundational control concepts are broadly applicable, current methods often operate in silos, targeting individual subsystems without accounting for the compositional or hierarchical nature of the full architecture.

Held in conjunction with a special session on the same topic, this workshop aims to seed a new interdisciplinary community centered on the study of control architectures and systems-level reasoning. Building on the success of the CDC 2024 and CDC 2025 editions (with over 100 researchers attending each), we will again convene researchers from robotics, control theory, applied mathematics, systems biology, neuroscience, and aerospace engineering to identify core challenges, share emerging methodologies, and develop a common language. A key outcome will be a collaboratively authored white paper outlining a long-term research agenda and identifying foundational problems, tools, and theoretical directions for layered and compositional control.

Speakers

Tentative Program

The workshop (full day, 8:45 am to 5:00 pm) will consist of a morning and afternoon session with talks from diverse speakers, concluding with lightning talks and a poster session from selected students and postdocs. Titles are tentative.

Time Speakers Title
08:45-09:00 Ames, Matni, Zardini Introduction to the workshop
09:00-09:30 Alessandro Abate (TBD) Learning and Verification of complex systems
09:30-10:00 James Anderson Architectures for Vision Based Control with World Models
10:00-10:30 Chris Verhoek and Nikolai Matni A Quantitative Framework for Navigating Controller Design Tradeoffs
10:30-11:00 Coffee break and discussions
11:00-11:30 Alberto Padoan Separation principles for optimization architectures
11:30-12:00 Matthew Hale Compositional Optimization for Layered Control Architectures
12:00-12:30 Dimitra Panagou Interconnecting Task and Motion Planning under Constraints and Uncertainty
12:30-14:00 Lunch break
14:00-14:30 Sylvia Herbert Composable safety monitors for long-horizon complex tasks
14:30-15:00 Runyu Zhang Structure-aware optimization for complex systems
15:00-15:30 Coffee break and discussions
15:30-16:00 Inigo Incer Designing Provably Safe and Secure Systems with Assume-Guarantee Contracts
16:00-onwards Selected Students & Postdocs Lightning Talks and Poster Session

Students and Postdocs Session

Students and postdocs will be given the opportunity to present. More info soon.

Previous Editions

We ran the first two editions of CAT at CDC 2024 and CDC 2025.