Comparison of Neural Network-Based Pose Estimation Approaches for Mobile Manipulation

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.

Online Dynamic Tip-Over Analysis for a Wheeled Mobile Dual-Arm Robot with an Improved Tip-Over Moment Stability Criterion

Tip-over stability analysis is critical for the success of mobile manipulation of the dual arm, especially in the cases that the dual arm or the mobile platform moves rapidly. Due to strong dynamics coupling between the dual arm and mobile platform, online evaluation of dynamic stability of the mobile dual-arm robot still remains challenging. This paper presents an improved tip-over moment stability criterion dealing with the dual arm and mobile platform interaction and proposes an algorithm for calculating the tip-over stability margin of the arm end in the workspace to analyze the dynamic stability of the wheeled mobile dual-arm robot. The simulations on a four-wheeled mobile dual-arm robot validate the correctness and feasibility of the proposed method.