Applicable scenario question

[biosbeta]

Your works look very impressive！However, from the test samples, the objects were all in a dispersed state of placement.Is the method you proposed suitable for objects with stacking and occlusion？

[pv33]

Them methods work for line-of-sight visible grasps. If the intent is to grasp an occluded object from the part that is visible, then it may/should work.

The method does not differentiate objects. If the goal is to grab an object on the top of the stack, then it should work, presuming that region is known. Often, when combined with an object recognition system or region of interest boundary, we can filter the set of grasps according to proximity or overlap to the region, then select from that filtered set.

If the objective is to grab an object that is visible and stack it, then it should be possible.