brackish labs
[ field sites ]

where we go to break things.

We prefer to publish site-level data and failure modes alongside the work, not just the highlight reel. These are the site types we're building access to.

  • intertidal mudflats

    scoping
    northern coast

    gait failure modes on soft, saturated substrate

    Tidal range of several meters, with hours-long windows of exposed mud. Our primary site type for studying foot-substrate interaction across a continuous wetness gradient.

  • rocky shorelines

    scoping
    open coast

    foothold planning on wave-wet rock

    Wet, irregular, biofouled surfaces. The hardest case for a vision-based foothold planner — high contrast, partial reflection, and a contact surface that changes friction across millimeters.

  • tidal marshes

    scoping
    estuarine

    locomotion through dense vegetation in standing water

    Vegetation hides the ground; the ground hides the water depth. A failure mode we expect to see often: confident gait into what looks like firm reed mat, ending in a sinkhole.

  • harbor edges

    scoping
    working ports

    perception against reflective, moving water surfaces

    Structured environment, anthropogenic clutter, constantly moving reflective surface. The dock face is a useful adversarial test for any visual stack we ship.

Site partnerships and access logs will be published as fieldwork begins.

Manage a coastal site that fits one of these types? We're actively looking for access partners.