RoboCup@Home — Service Robots in a Real House

RoboCup @Home takes robots somewhere nothing is predefined: a real person's house. The arena is an apartment — kitchen, living room, sofa, fridge, and people who may do something unexpected at any moment.

Three platforms

Open Platform League (OPL) — the team builds or chooses its own robot; maximum design freedom.
Domestic Standard Platform League (DSPL) — every team uses the standard Toyota HSR domestic robot.
Social Standard Platform League (SSPL) — using the Pepper social robot, weighted toward interaction rather than manipulation.

What the tests look like

Robots must do things that are trivial for us and catastrophic for them:

Receptionist: answer the door, ask the guest's name, introduce them to the room and find them a seat.
Storing groceries: take items from a table and shelve them next to similar categories.
Carry my luggage: follow one specific person through a crowded space carrying their bag.
General Purpose Service Robot (GPSR): hear a free-form spoken instruction and carry it out — the league's hardest test.
Restaurant: take and serve orders in an unfamiliar venue.

Why so hard? Because nothing is standard. Lighting differs, objects move, people speak ambiguously ("grab that one") and sometimes change their minds mid-task. The robot must handle ambiguity — a problem that never gets solved in a controlled lab.

Skills involved

Vision and object recognition in cluttered scenes
Natural language processing and spoken command understanding
Safe navigation among people and furniture
Manipulation — picking up a glass without breaking it
Social interaction — gaze, appropriate distance, polite response

With large language models, the command-understanding side has jumped forward, and @Home is now one of the best laboratories for combining language AI with physical robotics.

The student version

If the service-robot idea appeals but university-level @Home is out of reach, the FIRA Smart Home league implements the same idea at school scale with U14 and U19 brackets. Novin Zehn teams have taken first places there.

Frequently asked questions

Is @Home open to school students?

No, it is university level. The school-scale equivalent is the FIRA Smart Home league with U14 and U19 brackets.

Does the robot need an arm?

In OPL and DSPL, yes — there are pick-and-place tests. SSPL, using Pepper, is weighted toward social interaction.

How have language models changed this league?

Considerably. Understanding free-form, ambiguous human instructions was a long-standing bottleneck and has improved markedly with large language models.

Want to work on service robots?

Novin Zehn competition teams train from zero to the world stage — in Tehran and live online.