Irregular Depth Tiles: Automatically Generated Data Used for Network-based Robotic Grasping in 2D Dense Clutter

Recent advances in deep learning have enabled robots to grasp objects even in complex environments. However, a large amount of data is required to train the deep-learning network, which leads to a high cost in acquiring the learning data owing to the use of an actual robot or simulator. This paper presents a new form of grasp data that can be generated automatically to minimize the data-collection cost. The depth image is converted into simplified grasp data called an irregular depth tile that can be used to estimate the optimal grasp pose. Additionally, we propose a new grasping algorithm that employs different methods according to the amount of free space in the bounding box of the target object. This algorithm exhibited a significantly higher success rate than the existing grasping methods in grasping experiments in complex environments.

     

Localization using GPS and VISION aided INS with an image database and a network of a ground-based reference station in outdoor environments

GPS/INS integrated systems do not guarantee robustness and accuracy of localization, because GPS has vulnerability to external disturbances. However, the overall performance and reliability of the system can be significantly improved by fusing multiple sensors with a different operating principle. In outdoor environments where GPS may be blocked, there are many features compared to the open space and these features can provide much information for UGV localization. Thus, this paper proposes an improved localization algorithm based on the hierarchical federation of three measurement layers, i.e., GPS, INS, and visual localization, to overcome the shortcomings of GPS/INS integrated systems. The proposed algorithm automatically switches the operation modes according to GPS status and a network of a ground-based reference station. A vocabulary tree with SURF is used in the visual localization method. In the data fusion of visual localization and INS, an asynchronous and time-delayed data fusion algorithm is presented because visual localization is always time-delayed compared with INS. By using DGPS to obtain the reference position under the dynamic conditions of the reference station, the restrictions of the conventional DGPS are overcome and all UGVs within WiBro communication range of the reference station can accurately estimate the position with a common GPS. The experiment results with a predefined path demonstrate enhancement of the robustness and accuracy of localization in outdoor environments.     

Three-dimensional iterative closest point-based outdoor SLAM using terrain classification

To navigate in an unknown environment, a robot should build a model for the environment. For outdoor environments, an elevation map is used as the main world model. We considered the outdoor simultaneous localization and mapping (SLAM) method to build a global elevation map by matching local elevation maps. In this research, the iterative closest point (ICP) algorithm was used to match local elevation maps and estimate a robot pose. However, an alignment error is generated by the ICP algorithm due to false selection of corresponding points. Therefore, we propose a new method to classify environmental data into several groups, and to find the corresponding points correctly and improve the performance of the ICP algorithm. Different weights are assigned according to the classified groups because certain groups are very sensitive to the viewpoint of the robot. Three-dimensional (3-D) environmental data acquired by tilting a 2-D laser scanner are used to build local elevation maps and to classify each grid of the map. Experimental results in real environments show the increased accuracy of the proposed ICP-based matching and a reduction in matching time.     

Sonar Sensor-Based Efficient Exploration Method Using Sonar Salient Features and Several Gains

This paper describes a sonar sensor-based exploration method. To build an accurate map in an unknown environment during exploration, a simultaneous localization and mapping problem must be solved. Therefore, a new type of sonar feature called a ??sonar salient feature?? (SS-feature), is proposed for robust data association. The key concept of an SS-feature is to extract circle feature clouds on salient convex objects from environments by associating sets of sonar data. The SS-feature is used as an observation in the extended Kalman filter (EKF)-based SLAM framework. A suitable strategy is needed to efficiently explore the environment. We used utilities of driving cost, expected information about an unknown area, and localization quality. Through this strategy, the exploration method can greatly reduce behavior that leads a robot to explore a previously visited place, and thus shorten the exploration distance. A robot can select a favorable path for localization by localization gain during exploration. Thus, the robot can estimate its pose more robustly than other methods that do not consider localizability during exploration. This proposed exploration method was verified by various experiments, and it ensures that a robot can build an accurate map fully autonomously with sonar sensors in various home environments.     

Quantification of arm kinesthetic sense using an arm motiongenerator

The human response caused by the motion of an object grasped by a human operator is defined as an arm kinesthetic sense. Due to nonlinearity and ambiguity of human senses, there is no absolute standard for quantification of kinesthetic sense. In this research, a so-called two-dimensional (2-D) arm motion generator is developed to emulate various mechanical impedance, i.e., stiffness or damping, characteristics of a human arm. The words representing arm kinesthetic sense are selected and then the subject's satisfaction levels on these words for given impedance values are measured and processed by the semantic differential method and factor analysis. In addition, in order to reflect the individual differences of each subject in the arm kinesthetic sense, compensation for individual differences based on the neural network technique is proposed. Through this proposed algorithm, the human sensations to arm movements described qualitatively can be converted into engineering data ensuring objectivity, reproducibility, and universality. This database can be used to develop user-friendly products related to arm motion     

Real-time building of a thinning-based topological map

An accurate and compact map is essential to an autonomous mobile robot system. A topological map, one of the most popular map types, can be used to represent the environment in terms of discrete nodes with edges connecting them. It is usually constructed by Voronoi-like graphs, but in this paper the topological map is incrementally built based on the local grid map by using a thinning algorithm. This algorithm, when combined with the application of the C-obstacle, can easily extract only the meaningful topological information in real-time and is robust to environment change, because this map is extracted from a local grid map generated based on the Bayesian update formula. In this paper, position probability is defined to evaluate the quantitative reliability of the end node extracted by the thinning process. Since the thinning process builds only local topological maps, a global topological map should be constructed by merging local topological maps according to nodes with high position probability. For real and complex environments, experiments showed that the proposed map building method based on the thinning process can accurately build a local topological map in real-time, with which an accurate global topological map can be incrementally constructed.     

Proactive Human Search for the Designated Person with Prior Context Information in an Undiscovered Environments

This paper describes a scheme for proactive human search for a designated person in an undiscovered indoor environment without human operation or intervention. In designing and developing human identification with prior information a new approach that is robust to illumination and distance variations in the indoor environment is proposed. In addition, a substantial exploration method with an octree structure, suitable for path planning in an office configuration, is employed. All these functionalities are integrated in a message- and component-based architecture for the efficient integration and control of the system. This approach is demonstrated by succeeding human search in the challenging robot mission of the 2009 Robot Grand Challenge Contest.     

Complete and Rapid Regrasp Planning with Look-up Table

A pick-and-place operation in a 3-dimensional environment is a basic operation for humans and multi-purpose manipulators. However, there may be a difficult problem for such manipulators. Especially, if the object cannot be moved with a single grasp, regrasping, which can be a time-consuming process, should be carried out. Regrasping, given initial and final poses of the target object, is a construction of sequential transition of object poses that are compatible with two poses in the point of grasp configuration. This paper presents a novel approach for solving the regrasp problem. The approach consists of a preprocessing and a planning stage. Preprocessing, which is done only once for a given robot, generates a look-up table which has information on kinematically feasible task space of the end-effector throughout the entire workspace. Then, using this table, the planning automatically determines a possible intermediate location, pose and regrasp sequence leading from the pick-up to put-down grasp. With a redundant robot, it is shown experimentally that the presented method is complete in the entire workspace and can be implemented in real-time applications due to rapid regrasp planning time. The regrasp planner was combined with an existing path.     

Energy-based control of a haptic device using brakes.

This paper proposes an energy-based control method of a haptic device with electric brakes. Unsmooth motion is frequently observed in a haptic system using brakes during a wall-following task. Since it is generally known that a haptic system using brakes is passive due to brake's characteristics, its energy behavior has seldom been investigated. However, force distribution at the end effector reveals that the unsmooth motion of a haptic system using brakes represents active behavior of the system in the specific direction. A force control scheme is proposed that computes the gain for smooth motion by considering the energy behavior of a system. Experiments show that smooth wall following is possible with a proposed force control scheme.