Automatization for the picking and placing of a variety of objects stored on shelves is a challenging problem for robotic picking systems in distribution warehouses. Here, object recognition using image processing is especially effective at picking and placing a variety of objects. In this study, we propose an efficient method of object recognition based on object grasping position for picking robots. We use a convolutional neural network (CNN) that can achieve highly accurate object recognition. In typical CNN methods for object recognition, objects are recognized by using an image containing picking targets from which object regions suitable for grasping can be detected. However, these methods increase the computational cost because a large number of weight filters are convoluted with the whole image. The proposed method detects all graspable positions from an image as a first step. In the next step, it classifies an optimal grasping position by feeding an image of the local region at the grasping point to the CNN. By recognizing the grasping positions of the objects first, the computational cost is reduced because of the fewer convolutions of the CNN. Experimental results confirmed that the method can achieve highly accurate object recognition while decreasing the computational cost.
In this study, non-planar multi-rotor helicopter, which has individual inputs for six degrees of freedom (DOF) is proposed. Traditional multi-rotor helicopters with planar rotor arrangement are under-acutuated system, which has only four input for 6DOF. Therefore, translational and rotational motion occur simultaneously. This property of traditional multi-rotor helicopter is unfavorable on the inspection of infrastructure that is required to detect tiny cracks. In that case, attitude change results in shaking movie and image of camera on vehicle and missing the crack. Several mechanisms such as variable pitch mechanism, have been proposed to overcome that issue. However, in these cases, simplicity of the helicopter is drastically impaired. In this study, we propose the small unmanned aerial vehicle (SUAV) which can control 6DOF independently without additional actuator. To substantiate the effectiveness of that, we showed following two points: One of that is having equivalent flying performance. Another one is that is possible to move translational without attitude change.
Bin picking robot plays an important role in automated manufacturing in industries. One of the most important techniques of bin picking robot is object detection or pose estimation in bin scene, and recently various methods reported good performance. Some industrial parts have its 3D CAD model that can represent whole shape of those parts; hence these methods are based on comparing the model of target object with the scene measured by camera or 3D sensors. However, almost all foods handled at factory do not have fixed shape against the recent demand of automating pick and place task. Therefore model-based techniques cannot be applied. In this paper, we propose a method of object detection without its 3D CAD model by utilizing target surface area information and concavity of boundaries between respective individuals. We evaluate proposed method against some foods and show the effectiveness of the method.