The kick-off meeting of the Interreg SMART GUARD project was held in Labin on October 15–16. The meeting was hosted and organized by the City of Labin, the project’s lead partner, in cooperation with the Public Fire Brigade Labin, an associated partner.

Our team is a project partner responsible for providing scientific and technical expertise in simulation, analysis, and risk assessment related to wildfires and urban heat island phenomena.

More info on the project is avaliable at lab’s SMART GUARD project page.

Indago Lab is taking part in IROS 2026, the IEEE/RSJ International Conference on Intelligent Robots and Systems, held this year in Pittsburgh, USA, from September 27 to October 1. IROS is one of the major international robotics conferences, bringing together researchers and industry to present advances across robot perception, planning, control, learning, multi-robot systems, and autonomous systems.

This year, our work is represented by two papers focusing on autonomous multi-robot search.

The Department of Technical Sciences of the Croatian Academy of Sciences and Arts (HAZU), together with the Croatian Society of Mechanics (HDM), organized the Second Symposium on the Application of Artificial Intelligence in Computational Mechanics, which was held on September 18, 2026, in the HAZU Library Hall in Zagreb.

The lecture Data-driven flow dynamics in maritime search operations (Dinamika strujanja temeljena na podacima u pomorskim operacijama traženja) by Stefan Ivić and Karlo Jakac was presented at the symposium.

We are please to announce the start of Weave https://weave-research.net/ bilateral research project Diffusion-based Ergodic Coverage for Robotics (DECOR)

DECOR aims to study diffusion-based ergodic coverage approaches in robotics. The goal of this initiative is to advance ergodic control theory and its applications in robotic exploration, sensing, and manipulation. Ergodic control is an emerging paradigm in robotics that enables autonomous agents to explore spatial distributions in proportion to their informational density, producing natural and efficient search and coverage behaviors. Unlike conventional target-based control or stochastic sampling, ergodic control minimizes the mismatch between a robot’s time-averaged trajectory statistics and a target distribution, yielding robust performance in tasks such as active sensing, localization, surveillance, inspection, and manipulation under uncertainty.

We are happy to announce yet antoher seminar at our lab.

The seminar will take place on Tuesday, April 28, at 1:30 PM, as usual in room L10 (Computational Engineering Laboratory).

Presentations:

• Martina Ivić: Wind farm layout optimization
• Stefan Ivić: IndagoBench25 meta-optimization for determining best default parameters of optimization methods
• Siniša Družeta and Stefan Ivić: Support for integer, discrete, categorical, and periodic optimization variables in new Indago API

All interested researchers and students are warmly invited to attend. We look forward to your participation!

We are happy to announce the new release of Indago!

Check the PyPI for installation (https://pypi.org/project/Indago) and the documentation (https://indago.readthedocs.io/index.html)

New features

• New methods: Heap-Based Optimizer (HBO), Controlled Random Search (CRS)
• New samplers for generating initial population positions (Optimizer.sampler): Sobol (‘sobol’) and LatinHypercube (’lhs’)

Improvements

• IndagoBench25 results overview and corresponding guidelines added to documentation
• Template file for new methods included in the methods library (_new_method_template.py)
• Anakatabatic PSO variants are now using scipy.interpolate.make_interp_spline instead of legacy function interp1d
• Literature reference for MSGD provided in documentation
• LSHADE set as the default DE variant (performs much better than SHADE)
• Optimizer.scatter_method renamed to Optimizer.sampler

Bug fixes

• Fixed the problem of EFO crashing when using small pop_size values
• Using custom anakatabatic PSO without akb_model parameter now works without errors
• Fixed the error appearing when logging messages for some methods