Next-Best-View Artifact Scanning Robot

Scan Controller

Scanning…

Turntable speed 25°/s

Arm sweep speed 30°/s

Scanner field of view 22°

Scan Progress

Coverage

Points captured

0 / 0

Elevation bands

0 / 6

Scanner elevation

0°

How it works

This models a fixed-sensor, rotating-turntable digitization rig like the ones museums use to 3D-scan artifacts: a two-link robotic arm holds the scan head at a chosen elevation while the object spins underneath on a turntable, so one full revolution sweeps an entire ring of the surface.

Visibility test for surface point p:
  aim   = -scanner_pos (unit vector toward object centre)
  view  = normalize(p - scanner_pos)
  angle = acos(view · aim)
  toSrc = normalize(scanner_pos - p)
  visible  = angle < FOV/2  AND  (normal_p · toSrc) > 0.15

Next-best-view band choice (greedy set cover):
  band* = argmax over unvisited bands b of
          | { points assigned to b that are still unscanned } |

After each full turntable revolution, the controller marks the current elevation band done and jumps the arm to whichever remaining band still hides the most unscanned surface — the same greedy "next-best-view" heuristic real scan-planning software uses to minimise the number of viewpoints needed for full coverage.

The arm itself is driven by classic two-link inverse kinematics: given a fixed shoulder base, two rigid links of length L₁ and L₂, and a target position, the elbow angle follows from the law of cosines,

cos(θ_elbow) = (L1² + L2² − d²) / (2·L1·L2)

where d is the straight-line distance from the shoulder to the target. Every captured point is appended to a live point cloud attached to the artifact, so it keeps rotating with the object exactly like a real scan would.

Scanner / arm Captured point cloud

⚠ Couldn't load the 3D engineThree.js failed to load from the CDN. Check your connection and reload.