This paper proposes a novel method for randomized bin-picking based on learning. When a two-fingered gripper tries to pick an object from the pile, a finger often contacts a neighboring object. Even if a finger contacts a neighboring object, the target object will be successfully picked depending on the configuration of neighboring objects. In our proposed method, we use the visual information on neighboring objects to train the discriminator. Corresponding to a grasping posture of an object, the discriminator predicts whether or not the pick will be successful even if a finger contacts a neighboring object. We examine two learning algorithms, the linear support vector machine (SVM) and the random forest (RF) approaches. By using both methods, we demonstrate that the picking success rate is significantly higher than with conventional methods without learning.

Automating snap assemblies is highly desirable but challenging due to their varied geometrical configurations and elastic components. A key aspect to automating snap assemblies is robot state estimation and corrective motion generation, here defined as snap sensing. While progress is being made, there are yet no robust systems that allow for snap sensing. To this end we have integrated a framework that consists of a control strategy and control framework that generalises to cantilever snaps of varying geometrical complexity. We have also integrated a robot state verification method (RCBHT) that encodes FT data to yield high-level intuitive behaviours and perform output verification. Optimisation procedures and Bayesian filtering have been included in the RCBHT to increase robustness and granularity. The system provides belief states for higher level behaviours allowing probabilistic state estimation and outcome verification. In this work, preliminary assembly failure characterisation has been conducted and provides insights into assembly failure dynamics. The results, though still in simulation, are promising as the framework has effectively executed cantilever snap assemblies and robust robot state estimation with parts of varying complexity in two different robotic systems.

No abstract

No abstract

No abstract