Gesture recognition has multiple applications in medical and engineering fields. The problem of hand gesture consists identifying, at any moment, a given performed by the hand. In this work, we propose new model for real time. input is surface electromyography measured commercial sensor Myo armband placed on forearm. output label executed user proposed based Λ-nearest neighbor dynamic time warping algorithms. This can learn to recognize To evaluate performance our model, compared its...

This paper describes a method to intelligently schedule network of multiple RGBD sensors in Bayesian object tracking scenario, with special focus on Microsoft KinectTM devices. These setups have issues such as the large amount raw data generated by and interference caused overlapping fields view. The proposed algorithm addresses these selecting exclusively activating sensor that yields best measurement, defined novel stochastic model also considers hardware constraints intrinsic parameters....

This paper presents a novel approach to track nonconvex shape approximation of an extended target based on noisy point measurements. For this purpose, type random hypersurface model (RHM) called Level-set RHM is introduced that models the interior with level-sets implicit function. Based RHM, nonlinear measurement equation can be derived allows employ standard Gaussian state estimator for tracking object even in scenarios moderate noise. In paper, shapes are described using polygons, and...

Modeling 2D extended targets with star-convex Random Hypersurface Models (RHMs) allows for accurate object pose and shape estimation. A RHM models the of an aid a radial function that describes distance from center to any point on its boundary. However, up now only linear estimators, i.e., Kalman Filters, are used due lack explicit likelihood function. In this paper, we propose closed-form easy implement tracking RHMs. This makes it possible apply nonlinear estimators such as Particle...

In real world applications, robotic solutions remain impractical due to the challenges that arise in unknown and unstructured environments. To perform complex manipulation tasks cluttered situations, robots need be able identify interaction possibilities with scene, i.e. affordances of objects encountered. environments noisy perception, insufficient scene understanding limited prior knowledge, this is a challenging task. work, we present an approach for grasping scenes humanoid robot context...

We consider the task of recursively estimating pose and shape parameters 3D objects based on noisy point cloud measurements from their surface. focus whose surface can be constructed by transforming a plane curve, such as cylinder that is extruding circle. However, designing estimators for challenging, straightforward distance-minimizing approach cannot observe all parameters, additionally subject to bias in presence noise. In this article, we first discuss these issues then develop...

In this work, we introduce iviz, a mobile application for visualizing data in the Robot Operating System (ROS). last few years, popularity of ROS has grown enormously, making it standard platform robotic programming. However, availability environment is generally restricted to PCs with Linux operating system. Thus, users wanting see what happening system smartphone or tablet are stuck solutions such as screen mirroring web browser versions rviz, newer visualization modalities Augmented...

Despite significant advances in robot autonomy, manual intervention by a human operator is necessary many situations. This usually requires qualified staff and some robot-specific input device even for the comparatively simple case of platform locomotion. For this reason, we propose novel path generation method applicable to car-like vehicles. With method, “draws” desired 2D walking large-scale haptic interface while guiding force exerted, which ensures that generated can later be accurately...

As sensor resolution increases, the accuracy and robustness of tracking algorithms can be improved by incorporating more information about shape target object. This raises need for simple robust models capable describing detailed objects. In this paper we propose an approach based on Random Hypersurface Models that interprets shapes as scaled extrusions. is achieved combining projection-based with probabilistic approaches, integrating strengths both mechanisms. extruded such bottles, boxes,...

A popular approach when tracking extended objects with elongated shapes, such as ships or airplanes, is to approximate them a line segment. Despite its simple shape, the distribution of measurement sources on segment can be characterized in many radically different ways. The spectrum ranges from Spatial Distribution Models that assume distinct probability for each individual source, Greedy Association used curve fitting, which do not any at all. In between these border cases, Random...

In this paper, a novel measurement model based on spherical double Fourier series (DFS) for estimating the 3D shape of target concurrently with its kinematic state is introduced. Here, represented as star-convex radial function, decomposed DFS. comparison to ordinary DFS, DFS do not suffer from ambiguities at poles. Details will be given in paper. The representation integrated into Bayesian estimator framework via equation. As range sensors only generate measurements side facing sensor,...

In this paper, we propose a novel algorithm for automatically calibrating network of depth sensors, based on moving calibration object. The sensors may have non-overlapping fields view in order to avoid interference. Two major challenges are discussed. First, depending where the object is located relative sensor, number and quality measurements strongly varies. Second, single sensor observes only from one side. Dealing with these requires simple as well an that can deal under-determined...

As an alternative to Kalman filters and particle filters, recently the progressive Gaussian filter (PGF) was proposed for estimating state of discrete-time stochastic nonlinear dynamic systems. Like estimate PGF is a distribution, but like its measurement update works directly with likelihood function in order avoid inherent linearization filters. However, compared allows much faster estimation circumvents severe problem degeneracy by gradually transforming prior distribution into posterior...

In the past years, algorithms for 3D shape tracking using radial functions in spherical coordinates represented with different methods have been proposed. However, we seen that mainly measurements from lateral surface of target can be expected a lot dynamic scenarios and only few top bottom parts leading to an error-prone estimate regions when representation coordinates. We, therefore, propose represent function cylindrical coordinates, as these surface, no information or is needed. this...

In this paper, we propose a novel approach to track extended objects by incorporating negative information. While traditional techniques targets use only positive measurements, assumed stem from the target, proposed estimator is also capable of which tell us where target cannot be. To achieve this, introduce simple, robust, and easy-to-implement recursive Bayesian employs ideas field curve fitting. As an application idea, develop measurement equation estimate star-convex shapes can be used...

In this paper, we propose an algorithm for tracking mobile devices (such as smartphones, tablets, or smartglasses) in a known environment augmented reality applications. For purpose, interpret the extended object with shape, and design likelihoods different types of image features, using association models from tracking. Based on these likelihoods, together sensor information inertial measurement unit device, recursive Bayesian algorithm. We present results our first prototype discuss...