About me
The mind behind
IA Robots
Discover the origin, the vision and the work
that shapes the project.
Who am I?
I am Iker Alaminos, a Robotics Engineering graduate and founder of IA Robots. I work at the intersection of physics simulation, reinforcement learning and multi-agent systems, with a concrete focus: understanding how complex behaviours emerge in artificial systems when the environment, physics and objectives are well defined.
I am particularly interested in the "personality" that a robotic system acquires depending on how its training environment is designed. Two agents with the same architecture but different reward functions behave in radically different ways. For me, that is the robotics worth researching.
IA Robots is the way of documenting that work: a combination of rigorous engineering and open conceptual exploration, convinced that the future of robotics will not only lie in efficiency, but in understanding how far the autonomy of artificial systems can reach.
The two pillars of IA Robots
Space robotics and collective autonomy
Satellite swarms that learn to coordinate without explicit instructions, maintain formations under orbital perturbations and make distributed decisions when communication with the ground is impossible. In contexts where human intervention is minimal, this kind of autonomy is not optional — it is essential.
Conscious robotics and artificial cognition
What conditions are needed for complex cognitive processes to emerge in artificial systems? This is not about imitating human emotions, but about investigating whether concepts such as intention, identity or sensitivity can be emergent properties of the interaction between body, environment and learning.
The philosophy of IA Robots
IA Robots is built on a concrete working philosophy: combining the technical rigour of engineering with conceptual exploration that has no preset limits.
I am not only interested in systems that work correctly. I am interested in understanding why they work, what emerges from their interactions, and how far their autonomy can reach when the environment and the rules are well defined. Robotics is not just a discipline of solutions; it is a field of open questions.
The project is, in essence, a way of documenting that process: each experiment, each technical failure, each unexpected behaviour is valuable information. I seek a balance between the technical and the conceptual, between deep analysis and imagination, convinced that real progress comes from asking the hard questions before jumping to the answers.
The story behind the project
IA Robots was born from an early conviction: that the most interesting robots were not the ones that already existed, but the ones that did not yet exist. Not as replicas of industrial processes, but as systems capable of surprising even the people who design them.
Throughout the degree, every subject on control, simulation or artificial intelligence reinforced the same intuition. But it was when I started experimenting with reinforcement learning that something truly clicked: small changes in the environment or in the reward functions produced radically different behaviours. The agents were not following instructions; they were learning them. And sometimes, they learned things nobody had taught them.
That discovery became the backbone of the project and, later, the core of my Final Degree Project: a multi-agent system of autonomous satellites trained with reinforcement learning, validated first in simulation and then on real hardware.
IA Robots is the space where that process was documented. It is not the destination; it is the logbook.
What does it aim to convey?
"IA Robots aims to convey a vision of robotics that goes beyond efficiency: creating autonomous systems that not only solve problems, but learn, evolve and develop a more complex and meaningful form of interaction with their environment."
What am I working on now?
99.1% success in simulation. Lite6 validated on real hardware.
My Final Degree Project was completed with two parallel and independent experiments.
The first: a swarm of four autonomous satellites trained
with reinforcement learning, achieving 99.1% success under full orbital perturbations,
with distributed coordination and no inter-agent communication.
This system was fully validated in simulation.
The second: the UFactory Lite6 manipulator, designed as the type
of robotic arm that would be mounted on each satellite. Trained first in simulation
on an air bearings platform (planar microgravity) and then transferred
to the real physical robot in the laboratory, where the learned
policy worked correctly on hardware.
Two lines, one objective: to demonstrate that distributed robotic autonomy
is viable from learning all the way through to hardware.
If you want to talk about this…
If you are interested in what I am building, want to share ideas, ask questions or explore possible collaborations, I would be happy to hear from you.