In this paper, we present Sim-MEES: a largescale synthetic dataset that contains 1,550 objects with varying difficulty levels and physics properties, as well as 11 million grasp labels for mobile manipulators to plan grasps using different gripper modalities in cluttered environments. Our dataset generation process combines analytic models and dynamic simulations of the entire cluttered environment to provide accurate grasp labels. We provide a detailed study of our proposed labeling process for both parallel jaw grippers and suction cup grippers, comparing them with state-of-the-art methods to demonstrate how Sim-MEES can provide precise grasp labels in cluttered environments.

In this paper, we present two distinct neural network-based pose estimation approaches for mobile manipulation in factory environments. Synthetic datasets, unique to the factory setting, are created for neural network training in each approach. Approach I uses a CNN in conjunction with RBG and depth images. Approach II uses the DOPE network along with RGB images, CAD dimensions of the objects of interest, and the PnP algorithm. Each approach is evaluated and compared across pipeline complexity, dataset preparation resources, robustness, platform and run-time resources, and pose accuracy for manipulation planning. Finally, recommendations for when to use each method are provided.

This paper illustrates two approaches for mobile manipulation of factory robots using deep neural networks. The networks are trained using synthetic datasets unique to the factory environment. Approach I uses depth and RGB images of objects for its CNN and Approach II uses CAD models of the objects with RGB images for a DOPE network and PnP algorithm. Both the approaches are compared based on their complexity, required resources for training, robustness, pose estimation accuracy and run-time characteristics. Recommendations of which approach is suitable under what circumstances is provided. Finally, the most suitable approach is implemented on a real mobile factory robot in order to execute a series of manipulation tasks and validate the approach.