I research and build on robotic systems that have to deal with the real world — from learning and control to the physical world.
A few projects that show the kind of problems I like: physical systems, uncertainty, optimization and learning, with results that eventually have to survive contact with hardware.
I like the part of robotics where a mathematical idea has to become a behavior you can actually observe, measure and improve.
Dynamics, constraints, uncertainty, sensors, actuators — start with what the physical system will actually allow.
Use optimization, control, learning or perception where each is useful — without forcing one tool onto every problem.
Simulation is useful. Hardware is the judge. Experiments turn a promising idea into something you can trust.
Measure, debug, simplify, repeat. The interesting part usually starts when the first version doesn't work.
Research at the intersection of robotics, control and learning — with an emphasis on methods that can make sense on physical machines.
Uncertainty-aware trajectory optimization, coupled convex model predictive control and whole-body control for robust object transport with a quadruped.
A receding-horizon allocation strategy that accounts for asymmetric motor dynamics and exploits nullspace redundancy to smooth motor commands while preserving the desired wrench.
I’m Riccardo, a PhD student at Tampere University interested in how robotic systems can behave reliably in the messy physical world.
My work has taken me on control of aerial systems and quadruped robots, spanning both model-based control and reinforcement learning. I enjoy moving between theory, code and hardware — and learning something new when the problem demands it.