.. _point-tracking-pick-example:

Point tracking pick loop
=========================

Use calibrated RGB-D data to turn a caller-owned red-rectangle detection into a
robust 3D observation, resolve it into the planning frame, and apply a bounded
in-plane Cartesian correction. The example treats a side-on can as an
elongated red region and uses its short axis as the yaw channel in the same
re-observation loop.

Servoing starts from a fixed home pose with the tool pointing straight down,
and only ever corrects x, y, and yaw: height and the straight-down orientation
are held fixed throughout the loop, so it never drifts out of plane. Once
aligned, a separate pick step uses the last measured depth to descend, grasp,
and retreat back to the aligned pose.

.. tab-set::
    :sync-group: sdk-language

    .. tab-item:: Python
        :sync: python

        .. literalinclude:: scripts/point-tracking-pick.py
           :language: python

    .. tab-item:: TypeScript
        :sync: typescript

        .. literalinclude:: scripts/point-tracking-pick.ts
           :language: typescript