On January 15, the session Using Transient Controllers to Satisfy High-Level Multi-Robot Tasks took place, a DEEC TALK organized in partnership with the Institute for Systems and Robotics (ISR). The afternoon featured the participation of Dimos V. Dimarogonas, Professor in the Division of Decision and Control Systems, School of Electrical Engineering and Computer Science at KTH, and co-author of the book Formal Methods for Multi-Agent Feedback Control Systems. His research interests include multi-agent systems, hybrid systems and control, robotic navigation and manipulation, human–robot interaction, and networked control.

The DEEC TALK began with a brief introduction of the speaker by Sérgio Pequito, Vice-President for Research, Development, and External Relations.

Dimos V. Dimarogonas began by explaining that his motivation for developing transient controllers stems from the wide range of applications proposed in recent years, including multi-vehicle and multi-robot systems, in which he highlighted the use of drones and transport vehicles. However, despite the many developments and approaches that have emerged, the researcher emphasized that, from his perspective, the question of how to achieve and optimize immediate adaptation to the agents’ operating conditions, while maintaining system functionality, remains open.

Multi-robot systems are developed to address increasingly complex tasks. These tasks are distributed and involve communication among different agents, taking into account both spatial and temporal response parameters. Such parameters define actions through quantitative logics, such as Metric Interval Temporal Logic and Signal Temporal Logic.

As an example, the researcher described instructions given to a transport system: maintaining a distance of 1 meter from other robots (spatial parameter), monitoring an area periodically every 20 seconds (temporal parameter), and requesting coordination for collecting objects of interest and moving to the final station. In this way, the control and planning of each robot’s actions depend on formal logical verification languages.

In multi-robot systems, different robots may be assigned distinct but complementary tasks, reflecting a decentralized organizational logic. As a result, there is interdependence among agents both in terms of spatial positioning and action—due to required distances between robots and surrounding obstacles—and in temporal terms, such as timing, duration, and periodicity of actions. This dependence leads to an exponential increase in the number of states (such as robot position and task state) for each agent; in other words, system complexity grows as more robots are added.

For this reason, researchers resort to transient control methodologies, which focus on the system’s response during state changes, in order to develop solutions that account for both temporal and spatial parameters of each robot. These include decisions such as trajectory definition and timing of movement. Consequently, assigned tasks must be mathematically compatible with the available communication between robots—one of the key challenges in this field.

Throughout the session, Dimos V. Dimarogonas also shared the calculations used in the development of these systems, including contributions from several PhD students.

Currently, Dimos V. Dimarogonas’s research group is also applying these systems in aquatic environments.

The session concluded with questions from the students and researchers in attendance.