Eye tracking
Your gaze identifies what you want to reach.
Your gaze identifies what you want to reach.
Daily activity data helps anticipate what you might need.
A spoken response or deliberate blink confirms your choice.
The arm picks up the confirmed object and brings it within reach.
Connect the workspace, glasses, and wrist alignment view to follow the arm’s perspective during pickup.
Live camera feeds start when this view opens. Feeds are not recorded here.
Wrist alignment view, available during pickup
The arm, table, and objects in one view
Outward view from the glasses
Routine context is available when a Chalk source is connected.
No context connection details are available.
Explore a 3D desk model and the shuttle illustration, then return to the live camera views.
Loading the arm animations…
Real activity patterns to start. Your own choices over time.
A research dataset of everyday activities recorded in real homes: meals, sleep, drinking, phone use, and more.
Recorded from 2011 to 2023. Activities provide context; mapping them to specific objects is a design assumption.
As confirmed picks accumulate, your history matters more than the shared starting pattern.
Droid summarizes grab history and proposes hourly probabilities. Chalk uses deployed probabilities to rank likely objects alongside gaze.
Factory is not connected
Connect a routine-analysis service to view its recommendations.
Droid currently reads the synthetic grab baseline plus any logged real picks. CASAS provides a separate starting prior.
Compare gaze and history predictions from the earlier saved test run.
Accuracy and comparisons from the most recent completed evaluation.
No evaluation run yet.