Browse Bounties

Perform a series of fine motor tasks: thread a needle, button 3 different types of buttons (shirt, jacket, pant), and tie 2 different knots (shoelace knot and bow). Record with a closeup camera angle so finger movements, thread passage, and knot formation are clearly visible.

Walk across uneven outdoor terrain (grass, gravel path, dirt trail, or uneven pavement) while carrying objects. Show different carrying styles: in arms, on shoulder, in bag. Vary the weight and shape of objects. This captures balance, gait adaptation, and load-carrying behavior for robotics locomotion research.

Record yourself assembling a flat-pack furniture piece (IKEA-style bookshelf, desk, cabinet, or similar), then disassembling it back to flat. Capture all phases: unpacking, reading instructions, aligning pieces, inserting fasteners, and final adjustment. Then reverse the process for disassembly.

Use 5 different hand tools (hammer, screwdriver, wrench, pliers, tape measure, saw, chisel, level, etc.) and record each in use. Show realistic tool operation — grip, positioning, force application, and release. Capture tool choice behavior and handling patterns for robotic manipulation research.

Walk through a grocery store and select 5 items from different aisles. Record from a chest-mounted or handheld camera showing realistic retail navigation: reaching for items on shelves, reading labels, placing items in cart, and navigating between aisles.

Walk through your entire home from room to room, recording each space with a chest-mounted POV camera. Show realistic navigation behavior including opening doors, stepping over thresholds, turning corners, and moving furniture out of the way when needed. Cover every room in the home.

Fold a pile of laundry — minimum 10 items including shirts, pants, socks, and towels. Record from a front-facing or overhead angle so hand movements, folding techniques, and garment handling are clearly visible. Show different folding methods for different garment types.

Pick up and set down 15 different objects from a realistically cluttered desk surface. Objects include: pen, phone, notebook, coffee mug, keys, USB drive, paperclip, earbuds, bottle, sticker sheet, tape dispenser, sticky notes, card, glasses, and one object of your choice. Three takes, different object orderings each time.

Record yourself organizing kitchen cabinets and drawers — sorting items, grouping by category, wiping surfaces, and realigning containers. This captures the precise hand movements and spatial reasoning involved in domestic organization tasks.

Film yourself preparing a simple meal from start to finish using a POV chest-mounted camera. Show realistic kitchen behavior including ingredient handling, tool use, and cooking techniques. This data helps robotics researchers study kitchen manipulation and safe human-robot collaboration in home environments.

Record a 15-minute outdoor gardening session covering planting seedlings, pulling weeds, turning soil with a trowel or hand fork, and watering. We need diverse outdoor conditions — different soil textures, varying plant sizes, real garden clutter. This data trains robotic systems for unstructured outdoor manipulation tasks.

Record yourself replacing a P-trap or compression fitting under a kitchen or bathroom sink. We need real repair scenarios showing the full workflow: inspecting the problem, shutting off water, removing old parts, installing new ones, and testing. Authentic repair environments with actual corrosion or wear preferred — not a clean demo.

Record your natural shopping behavior in a retail store (grocery, pharmacy, general merchandise). Capture browsing shelves, reading labels, comparing products, reaching for items, and placing them in a cart/basket. We need authentic shopping behavior, not staged.

Record natural conversations between two bilingual speakers who naturally code-switch between Spanish and English. We need authentic, unscripted dialogue covering everyday topics. Both speakers should be comfortable switching languages mid-sentence as they would naturally.

Seeking LEFT-HANDED individuals to record themselves opening various jars, bottles, and containers, then using common hand tools (screwdrivers, wrenches, scissors). This data specifically addresses the underrepresentation of left-handed manipulation in existing training datasets. Must be performed with left hand as dominant.

Record a typical 10-minute segment of desk work including typing on a keyboard, handling papers/documents, organizing desk items, and using a mouse. We are training AI models to understand routine office workflows from an egocentric perspective. Real office or home office environment required.

Record yourself walking through various indoor spaces with obstacles and clutter: hallways with boxes, living rooms with furniture, storage rooms, etc. We need natural navigation behavior including turning, stepping over/around objects, and adjusting path. Multiple rooms and transitions preferred.

Record yourself preparing a full meal from scratch in your home kitchen. We need diverse footage of chopping, stirring, reaching for ingredients, using appliances, and plating. Must include at least 3 different cooking techniques (e.g., chopping, sauteing, mixing). Natural kitchen environment with typical clutter is preferred over a studio setup.