Abstract: An autonomous mobile robot that is equipped with functionalities to assist the elderly and disabled patients to live at home in a way that is acceptable and desirable for the patients and caregivers is described. The robot provides safety monitoring, cognitive and communication support to patients. mobility to ensure availability, and a scalable platform. The robot is able to detect when the robot has toppled over and automatically execute operations that restore the robot to a full upright position.

Abstract: According to one example for capturing three-dimensional data of an object, illumination of an object is initiated, the object is scanned with a sensor, and a 3D representation of the object is created. Illumination of the object may comprise illumination of a surface of the object within a field of view of the sensor.

Abstract: Processes and systems are directed to training a neural network of an object recognition system. The processes and systems record video streams of people. Sequences of object images are extracted from each video stream, each sequence of object images corresponding to one of the people. A triplet comprising an anchor feature vector and a positive feature vector of the same object and a negative feature vector of a different object of feature vectors are formed for each sequence of object images. The anchor, positive, and negative feature vectors of each triplet are separately input to the neural network to compute corresponding output anchor, positive, and negative vectors. A triplet loss function value computed from the output anchor, positive, and negative vectors. When the triplite loss function value is greater than a threshold, the neural network is retrained using the anchor and positive feature vectors of the sequences of object images.

Abstract: Processes and systems are directed to training a neural network of an object recognition system. The processes and systems record video streams of people. Sequences of object images are extracted from each video stream, each sequence of object images corresponding to one of the people. A triplet comprising an anchor feature vector and a positive feature vector of the same object and a negative feature vector of a different object of feature vectors are formed for each sequence of object images. The anchor, positive, and negative feature vectors of each triplet are separately input to the neural network to compute corresponding output anchor, positive, and negative vectors. A triplet loss function value computed from the output anchor, positive, and negative vectors. When the triplite loss function value is greater than a threshold, the neural network is retrained using the anchor and positive feature vectors of the sequences of object images.

Abstract: A method includes receiving data representing an image captured of an object disposed on a surface in the presence of illumination by a flash light. The technique includes processing the data to identify an object type associated with the object and further processing the data based at least in part on the identified object type.

Abstract: Examples disclosed herein relate to detecting misalignment of a touch sensitive mat. Examples include detecting corners of the touch sensitive mat, determining a set of reference corners, performing a comparison of the detected corners of the mat with the set of reference corners, and determining a level of misalignment based on the comparison.

Abstract: Examples disclosed herein relate to a mat characteristic to process images. Examples include to acquire an image of a mat from a camera in a computing device; to process the image according to the mat characteristic in the computing device; and to display the processed image.

Abstract: According to one example for outputting image data, an image comprising a surface and an object are captured on a sensor. An object mask based on the captured image is created on a processor. A first composite image based on the object mask and a source content file is created. In an example, the first composite image is projected to the surface.

Abstract: Examples disclosed herein relate to a mat characteristic to process images. Examples include to acquire an image of a mat from a camera in a computing device; to process the image according to the mat characteristic in the computing device; and to display the processed image.

Abstract: Examples disclosed herein relate to a mat characteristic to process images. Examples include to acquire an image of a mat from a camera in a computing device; to process the image according to the mat characteristic in the computing device; and to display the processed image.

Abstract: Examples disclosed herein relate to identifying a target touch region of a touch-sensitive surface based on an image. Examples include a touch input detected at a location of a touch-sensitive surface, an image representing an object disposed between a camera that captures the image and the touch-sensitive surface, identifying a target touch region of a touch-sensitive surface based on an image, and rejecting the detected touch input when the location of the detected touch input is not within any of the at least one identified target touch region of the touch-sensitive surface.

Abstract: An example processor-implemented method for generating corners of a display area is provided. The method comprises detecting a dominant line for each side of the display area, each dominant line used to identify corners, detecting subline segments on each side of the display area, determining a distance between the corners identified by the dominant lines and the sub-line segments on each side, and generating the corners of the display area based on the distance.

Abstract: A system includes a sensor to capture multiple images of a portion of a first object illuminated by coherent illumination and a time of capture of each of the images; and a processor to compare two images of the multiple images to identify one or more touch points. Each touch point has a difference in value between the two images that is greater than a threshold. Upon determining a spatial shape formed by the identified touch points that corresponds to a pointing end of a pointing object, the system provides at least one of: i) a touch location of the pointing end relative to the first object, where the touch location is based on the spatial shape formed by the identified touch points, or ii) the time of capture of a second image of the two images that produced the spatial shape.

Abstract: In some examples, a computing device processes depth images to capture an interaction of an object relative to an interaction plane.

Abstract: Examples disclosed herein relate to aligning content displayed from a projector on to a touch sensitive mat. Examples include detecting a border of the mat, wherein the mat includes a surface area of a first spectral reflectance characteristic on to which the projector is to project the content, and the border of a second spectral reflectance characteristic different from the first spectral reflectance characteristic surrounding a perimeter of the surface area. As an example, detecting the border of the mat generally includes differentiating the second spectral reflectance characteristic of the border from the first spectral reflectance characteristic of the surface area. Examples include detecting a border of the content displayed on to the mat, and adjusting projector settings for the border of the content displayed on to the mat to fit within the detected border of the mat.

Abstract: According to one example for segmenting image data, image data comprising color pixel data, IR data, and depth data is received from a sensor. The image data is segmented into a first list of objects based on at least one computed feature of the image data. At least one object type is determined for at least one object in the first list of objects. The segmentation of the first list of objects is refined into a second list of objects based on the at least one object type. In an example, the second list of objects is output.

Abstract: Examples disclosed herein relate to determining a segmentation boundary based on images representing an object. Examples include an IR image based on IR light reflected by an object disposed between an IR camera and an IR-absorbing surface, a color image representing the object disposed between the color camera and the IR-absorbing surface, and determining a segmentation boundary for the object.

Abstract: Examples disclosed herein relate to determining a segmentation boundary based on an image representing an object. Examples include capturing, from an infrared (IR) camera disposed above and pointed at a projection screen, an IR image representing an object disposed between the IR camera and the projection screen, based on an intensity of IR light reflected by the object and the projection screen including a surface to specularly reflect IR light. Examples include determining a segmentation boundary based at least in part on at least one boundary between a first set of IR light intensity values and a second set of IR light intensity values, wherein the segmentation boundary represents at least one outer edge of the object based on the IR image.

Abstract: In some examples, a computing device processes depth images to capture an interaction of an object relative to an interaction plane.

Abstract: According to one example for outputting image data, an image comprising a surface and an object are captured on a sensor. An object mask based on the captured image is created on a processor. A first composite image based on the object mask and a source content file is created. In an example, the first composite image is projected to the surface.