SimReady assets: appliances
Washers, dryers, dishwashers, fridges, microwaves, vacuums — heavy, articulated, and full of door and control surfaces a home robot has to recognise, reach, and interact with.
Home robots interact with appliances constantly: opening a fridge, loading a dishwasher, starting a washing machine, retrieving food from a microwave. Each appliance is a heavy stationary obstacle with articulated doors, drawers, control panels, and the occasional rotating drum. Rigyd produces SimReady OpenUSD and MJCF with calibrated mass, accurate door articulation, and the surface-friction tuning a home-robot policy needs to transfer from sim to the real kitchen or laundry room.
What's in this category
Refrigerators and freezers
Hinged doors, internal shelves and drawers, magnetic seals. Common manipulation benchmark.
Washing machines and dryers
Front-loaders with circular doors; top-loaders with hinged lids. Drum visible from outside.
Dishwashers
Drop-down doors, sliding internal racks, detergent dispensers. Multi-step loading task.
Microwaves and ovens
Counter or built-in, hinged or drop-down doors, turntables, control panels.
Coffee machines
Hopper lids, drip trays, removable water tanks. Surfaces of mixed plastic and metal.
Vacuum cleaners
Upright, canister, robot vacuums. The latter are themselves robotic obstacles for other robots.
Toasters and small countertop appliances
Slot mechanisms, sliding levers, dial controls. Standard manipulation targets.
Range hoods and exhaust fans
Ceiling-mounted, articulated filters, control panels. Reach-planning challenges.
Air purifiers and humidifiers
Removable tanks, replaceable filters. Periodic-maintenance task targets.
Physics characteristics
Heavy mass, stable footprint
Major appliances (fridges 60–150kg, washers 60–90kg, dishwashers 40–60kg) are heavy stationary obstacles. Rigyd computes mass from volume × material density with calibration tuned for the steel + insulation composite construction of major appliances. Tip-over is rare under normal robot interaction but matters for safety testing.
Articulated doors, drawers, and panels
Every appliance has at least one articulated surface — usually a door or lid, often multiple drawers or sliding racks. Rigyd outputs OpenUSD with PhysicsArticulationRootAPI for each appliance with revolute joints for hinged doors and prismatic joints for sliding racks. MJCF output uses native MuJoCo joint declarations.
Mixed-material surfaces and controls
Appliance fronts are typically stainless steel, painted metal, or molded plastic. Friction coefficients vary materially. Control panels (buttons, dials, touch-screens) are separate prims with their own contact behaviour. Rigyd keys per-surface friction to material classification so robot-finger contact on a button differs from a sliding hand on the door.
Common materials
Robot tasks these assets enable
Appliance operation
Opening doors, pressing buttons, turning dials, loading and unloading. The bulk of home-robot interaction is appliance operation; correct articulation and control-surface geometry drive whether the policy transfers.
Indoor navigation around appliances
Robot vacuums, delivery robots, and humanoid platforms navigate kitchens and laundry rooms full of appliances. Door-open states change the navigable space; an open dishwasher door is a navigation obstacle a closed one is not.
Maintenance and repair tasks
Future appliance-repair robots need to recognise, reach, and disassemble appliances. Rigyd outputs preserve articulation hierarchy and component identity from the source CAD, which is the foundation a repair policy needs.
Compatible simulators
The same source asset feeds every simulator below from one Rigyd output.
Related robotics verticals
~5 min
per appliance, including articulation
per-surface
friction tuned by material classification
OpenUSD + MJCF
usable in Isaac Sim, MuJoCo, Genesis from one source
FAQ
How does Rigyd handle articulated doors and drawers on appliances?
Articulation hierarchy is detected from the source geometry — separate door, drawer, and rack meshes are recognised as articulated children. Rigyd applies PhysicsArticulationRootAPI on the appliance body with revolute joints for hinged doors, prismatic joints for sliding racks, and configurable joint friction so a magnetic-seal fridge door behaves differently from a free-swinging cabinet door. MJCF uses MuJoCo native joints with the same kinematics.
Can the assets simulate door magnetic seals or detents?
Door magnetic seals and detents are modelled as additional joint forces in the simulator (not as a separate physics primitive in the asset). The Rigyd output exposes the joint with default friction and damping; for accurate magnetic-seal behaviour, scripts or simulator extensions apply a position-dependent torque at the closed position. Most home-robot training tolerates the default joint behaviour; magnetic-seal accuracy matters only for tasks specifically about door interaction force.
Are control panels (buttons, dials, touch-screens) reachable manipulation targets?
Yes. Control surfaces are emitted as separate prims with their own collision geometry and friction. Buttons can be modelled as small prismatic joints if push-button interaction needs to be simulated; dials as revolute joints. Touch-screens are flat collision surfaces with the screen geometry visible for vision training. The level of detail is tunable per asset class.
Does Rigyd model rotating drums or moving parts inside running appliances?
Rigyd produces the static and articulated geometry — the drum is a separately identified prim with the correct mass and inertia. Driving the drum rotation as part of an appliance run (e.g. a washing-machine cycle) happens at simulation time using simulator-side scripting against the joint. The asset gives you the parts; the operating cycle is configured per training scenario.
Which simulators consume these appliances for home-robot training?
OpenUSD output drops into NVIDIA Isaac Sim, Isaac Lab, Omniverse, Unreal Engine, and Unity, plus Gazebo Sim via USD imports. MJCF output drops into MuJoCo, MJX, and Genesis. Home-robot training pipelines that combine GPU-parallel rollout (Isaac Lab) with precise contact-rich evaluation (MuJoCo) get both stages from one Rigyd source asset, which is the practical reason teams switch from manual asset prep to automated.
Train home robots to operate the appliances actually in the home
Drop in CAD, images, or text descriptions and get SimReady appliances in minutes per unit.
Request API Access