Joint Learning of Instance and Semantic Segmentation for Robotic Pick-and-Place with Heavy Occlusions in Clutter
This work addresses robotic manipulation challenges in cluttered settings, though it appears incremental by building on existing instance segmentation models.
The paper tackles the problem of robotic pick-and-place in cluttered environments with heavy occlusions by proposing a joint learning approach for instance and semantic segmentation, resulting in improved performance over instance-only methods as evaluated on test datasets and applied to real-world robotic tasks.
We present joint learning of instance and semantic segmentation for visible and occluded region masks. Sharing the feature extractor with instance occlusion segmentation, we introduce semantic occlusion segmentation into the instance segmentation model. This joint learning fuses the instance- and image-level reasoning of the mask prediction on the different segmentation tasks, which was missing in the previous work of learning instance segmentation only (instance-only). In the experiments, we evaluated the proposed joint learning comparing the instance-only learning on the test dataset. We also applied the joint learning model to 2 different types of robotic pick-and-place tasks (random and target picking) and evaluated its effectiveness to achieve real-world robotic tasks.