This paper proposes an unknown object detection technique for mobile robots that can be applied in indoor environments with floors including several non-horizontal partial areas. In these environments, practical techniques using a 2D range finder with an assumption of robots' horizontal locomotion cannot distinguish between unknown objects and floors when robots' sensors slightly rotate to downward direction on small slopes. On the other hand, conventional techniques with 3D sensing need several orders of magnitude higher computational costs and memory usages compared with practical 2D techniques. To address these issues, the proposed technique introduces a floor height map and a 3D gyroscope in addition to a conventional 2D object map and a 2D range finder. By registration between measured range data and the two types of 2D maps using 3D orientation obtained from the gyroscope, unknown objects can be detected and distinguished from floors with computational costs and memory usages in the same order of magnitude as practical 2D techniques. In addition, using the floor height map, the proposed technique can also collect the 3D orientation that has accumulative errors caused by integration of angular velocities measured by the gyroscope. The function can improve sustainability and practical utility of environment recognition including unknown object detection.
This paper describes the liquid container transfer by robot, due to the demand for the conveyance of molten metal in casting processes, for the tray service by robot, and so on. Since a liquid in container can wave greatly by changes of transferring velocity due to the inertial force acting on its own, robots cannot carry the container as a rigid box. Therefore, transferring liquid containers using robots requires a carrying method which prevents content fluid from spilling from the container. The proposed method, TGCW (Trajectory Generation to Control Wave) method, is a feed-forward control to realize the prevention of the spillover of content fluid. To reveal the performance of the TGCW method, we performed the verification simulations using the Explicit-MPS (Moving Particle Simulation) method and the validation experiments using a robot arm. We obtained the result that the proposed method has the effect of reducing the displacement of liquid and the simulation results close to experiments.
This paper describes an underactuated robot finger whose link posture is controlled by using dynamical effect. The finger posture control method where the vibration center angles of the active joint and the passive one are controlled independently of each other, is proposed. First we introduce the analytical model of two-joint two-link system which has the active joint driven by an actuator and the passive joint with an elastic element. Based on the model, we theoretically reveal Variable Vibration Center Effect (VVCE) of the elastic joint with respect to the input frequency of the active joint. We then discuss how to control the finger link posture by utilizing the effect. After showing the finger mechanism design, we experimentally confirm the proposed method and demonstrate a grasping motion by the prototype.
A method to detect hazards in the sense of traversability on pedestrian paths is presented for the purpose of supporting hazard recognition by robotic transporter users. Target transporters are supposed to be equipped with a 3D range image sensor to observe potential hazards like road roughness, drivability, and positive/negative obstacles. Hazard detection problem is formulated as a supervised learning such that point cloud data sets labeled hazard or non-hazard by a transporter user are input to the learning system and the learned output produces a hazard prediction to a novel data set. The present study tackles a reliable feature selection from a clustered point cloud with observation density inversely proportional to the square of the distance and found that feature selection by mean shift clustering works better than those by graph based clustering. Experiment results show that the proposed method is able to discriminate hazard cases from non-hazard ones of traffic control posts, walls, and down steps along pedestrian paths that previous approaches fail to predict correctly. The processing time of those predictions meets the requirements for online hazard detection.