Zach Christensen
Mimic
Stage 3Sessions 39 to 45

Eyes and Voice

A calibrated camera, a marker's pose resolved in the arm's own frame, and a spoken instruction that gets validated before anything moves. The rule the whole stage is built on: the model proposes, and my code decides.

Done when
An arm that sees and listens.
Progress
0 of 7 sessions0%

Sessions 39 to 45

Eyes and voice

A camera calibrated properly, a marker's pose in the arm's own frame, and a spoken instruction validated before anything moves.

  1. Not built yet

    Camera, calibrated

    Explain what an intrinsic matrix and a distortion model each describe, and quote a reprojection error and say what it means.

  2. Not built yet

    ArUco markers before neural networks

    Explain what PnP solves and why four known corners in one calibrated image recover a full six-DOF pose.

  3. Not built yet

    Hand-eye calibration

    Explain what AX = XB is asking, and tell a calibration error from a noise problem by the shape of the error.

  4. Not built yet

    Pick and place, closed loop

    Enumerate a pipeline's failure modes before writing its happy path, and say what each one should do.

  5. Not built yet

    Local speech to text

    Measure a speech pipeline rather than demonstrating it, and quote a word error rate taken in the room you actually use.

  6. Not built yet

    Language to intent

    State why a model's output must be validated before it reaches anything that moves, and express that as a schema with bounds.

  7. Not built yet

    Speak back

    Explain why a refusal is worth more than a confirmation in assistive technology, and keep the phrases as data rather than code.