Robotic Grasping Technology Based on Shape Analysis and Probabilistic Reasoning

In the task of grasping irregular objects, the transported objects may shake and fall off due to their complex and diverse shapes and structures. For these issues, a robotic grasping technology based on shape analysis and probabilistic reasoning is proposed. Firstly, the dispersivity and flatness of the object's point cloud are analyzed to generate a set of candidate grasping poses. Then, the factors influencing the shaking and falling off of the object are qualitatively analyzed in the simulation scenario, and the number of successful grasping and rotation-translation experiments is statistically counted in the simulation. The stability of the grasp pose is quantitatively analyzed using the conditional expectation method, and a PointNet discriminator is trained to evaluate and rank the candidate grasp poses. The grasping is ultimately completed with the optimal grasp pose. The experimental results indicate that the proposed method can solve the issue of shaking and falling off of irregular objects during the grasping and transporting process. Compared with the benchmark method, the average grasping success rate is improved to 89.2%, an increase of 2.6%, and the average transportation stability is enhanced to 84.2%, an increase of 22.7%. The proposed method enables intelligent grasping of objects in multi-object stacking scenarios, ensuring stability during the grasping and transporting process, and establishing a logical sequence for grasping.