2D fits
Fixtures, trays, conveyors with limited height variation; find–orient–correct in-plane.
Knowledge · Robotics
2D vs 3D robot vision: when is 2D enough?
Stable presentation and planar correction: often 2D. Free height, nesting or grasps in a bin: 3D or bin picking. The scene decides, not the hype.
Decision frame
Look at presentation variation
If X/Y/Rz in-plane suffice and Z/tilt are predictable, 2D is usually faster to validate. Once Z or 3D orientation drives the grasp, step up.
3D guidance fits
Shallow nests, shape variation, height that affects the gripper, 6D pose.
Bin picking is separate
Messy bins with occlusion and collision: own workflow (point cloud → grasp).
Pitfalls
Why ‘just use 3D’ is not automatically better
More data means more calibration, compute and failure modes. Feasibility on real parts remains mandatory.
Lighting first
Poor contrast or glare breaks both 2D and 3D.
Hand-eye
Without stable calibration, pose accuracy is marketing.
Takt
Dense 3D every cycle can be too heavy, ROI or 2.5D may suffice.
Keep reading
Solution path
Torn between 2D and 3D?
Send a photo of presentation and gripper, we say which path is worth testing.