Patent Application
20190384406
The present disclosure generally relates to Information Handling Systems (IHSs), and, more particularly, to systems and methods for enabling one-handed gesture sequences in virtual, augmented, and mixed reality (xR) applications.
As the value and use of information continues to increase, individuals and businesses seek additional ways to process and store information. One option available to users is Information Handling Systems (IHSs). An IHS generally processes, compiles, stores, and/or communicates information or data for business, personal, or other purposes thereby allowing users to take advantage of the value of the information. Because technology and information handling needs and requirements vary between different users or applications, IHSs may also vary regarding what information is handled, how the information is handled, how much information is processed, stored, or communicated, and how quickly and efficiently the information may be processed, stored, or communicated. The variations in IHSs allow for IHSs to be general or configured for a specific user or specific use such as financial transaction processing, airline reservations, enterprise data storage, or global communications. In addition, IHSs may include a variety of hardware and software components that may be configured to process, store, and communicate information and may include one or more computer systems, data storage systems, and networking systems.
IHSs may be used to produce virtual, augmented, or mixed reality (xR) applications. The goal of virtual reality (VR) is to immerse users in virtual environments. A conventional VR device obscures a user's real-world surroundings, such that only digitally-generated images remain visible. In contrast, augmented reality (AR) and mixed reality (MR) operate by overlaying digitally-generated content or entities (e.g., characters, text, hyperlinks, images, graphics, etc.) upon the user's real-world, physical surroundings. A typical AR/MR device includes a projection-based optical system that displays content on a translucent or transparent surface of an HMD, heads-up display (HUD), eyeglasses, or the like (collectively “HMDs”).
In various implementations, HMDs may be tethered to an external or host IHS. Most HMDs do not have as much processing capability as the host IHS, so the host IHS is used to generate the digital images to be displayed by the HMD. The HMD transmits information to the host IHS regarding the state of the user, which in turn enables the host IHS to determine which image or frame to show to the user next, and from which perspective, as the user moves in space.
Embodiments of systems and methods for enabling one-handed gesture sequences in virtual, augmented, and mixed reality (xR) applications are described. In an illustrative, non-limiting embodiment, an Information Handling System (IHS) may include a processor and a memory coupled to the processor, the memory having program instructions stored thereon that, upon execution, cause the IHS to: receive a gesture sequence from a user wearing a Head-Mounted Device (HMD) coupled to the IHS, where the HMD is configured to display an xR application; identify the gesture sequence as a one-handed sequence; and execute a command in response to the one-handed sequence.
To receive the gesture sequence, the program instructions, upon execution, may cause the IHS to receive a plurality of video frames captured using a camera mounted on the HMD. To identify the one-handed sequence, the program instructions, upon execution, may cause the IHS to perform a gesture recognition process upon: a first set of video frames corresponding to a start phase, a second set of video frames corresponding to a motion phase, and a third set of video frames corresponding to an end phase.
For example, the one-handed sequence may be a select gesture sequence including a hand with palm out hovering over an object to be selected with fingers spread apart over the object, and the command may add a visual effect to at least a portion of the object. The select gesture sequence may also include index finger movement over a given portion of the object, and adding the visual effect may include adding the visual effect to the given portion of the object to the exclusion of a remainder portion of the object. Additionally, or alternatively, the select gesture sequence may include an annotation gesture, and the program instructions, upon execution, may cause the IHS to enable the user to associate an annotation with the given portion of the object. Moreover, a deselect gesture sequence may include: (i) a hand with palm out hovering over an object with fingers spread apart followed by (ii) fingers being closed together over the object; and the command may remove a visual effect from the object.
In some cases, the one-handed sequence may be a menu open gesture sequence including a hand held up with palm facing in and fingers together, and the command may display a menu over the hand. The menu open gesture sequence may include at least one of: (i) the hand moving away while a visual effect is displayed around the menu, and the command closes the menu; or (ii) the hand moving away after the visual effect disappears, and the command maintains the menu on display. Moreover, a one-handed sequence may be a menu close gesture sequence including: (i) a hand held up with a palm facing in behind a menu until the menu is selected, followed by (ii) fingers closing; and the command closes the menu.
The one-handed sequence may also be a menu repositioning gesture sequence including: (i) a hand held up with a palm facing in behind a menu until the menu is selected, followed by (ii) the hand moving to another position with the palm facing in; and the command may move the menu to the other position.
Additionally, or alternatively, the one-handed sequence may be a workspace minimization gesture sequence including: (i) a hand horizontally leveled with palm facing down at a top or bottom portion of a display, followed by (ii) the hand moving horizontally leveled with palm facing down to an opposite portion of the display across a workspace; and the command may minimize the workspace and display a placeholder. Additionally, or alternatively, the one-handed sequence may be a workspace maximization gesture sequence including: (i) a hand horizontally leveled with palm facing down at a top or bottom portion of a display having a placeholder, followed by (ii) the hand moving horizontally leveled with palm facing down to an opposite portion of the display; and the command may maximize the workspace and remove the placeholder.
Additionally, or alternatively, the one-handed sequence may be a lock gesture sequence including: (i) a hand hovering over an object to be locked with index and middle fingers in extended position, and (ii) the index and middle fingers crossed; and the command may lock access to the object. Additionally, or alternatively, the one-handed sequence may be an unlock gesture sequence including: (i) a hand hovering over an object to be locked with index and middle fingers crossed, followed by (ii) the index and middle fingers in extended position; and the command may unlock access to the object. Additionally, or alternatively, the one-handed sequence may be a collaborator list gesture sequence including: a hand moving four extended fingers, and the command may display a list of collaborators in the xR application.
In some implementations, to receive the gesture sequence, the program instructions, upon execution, may cause the IHS to receive an audio signal captured using a microphone mounted on the HMD. In that case, for example, the one-handed sequence may be an audio mute or unmute gesture sequence including: at least one hand covering or uncovering at least one ear, and the command may mute or unmute only one audio channel.
In an another illustrative, non-limiting embodiment, a method may include: receiving a gesture sequence from a user wearing an HMD coupled to the IHS, wherein the HMD is configured to display an xR application; identifying the gesture sequence as a one-handed sequence; and executing a command in response to the one-handed sequence. In yet another illustrative, non-limiting embodiment, a hardware memory device of an IHS may have program instructions stored thereon that, upon execution by a hardware processor, cause the IHS to: receive a gesture sequence from a user wearing an HMD coupled to the IHS, wherein the HMD is configured to display an xR application; identify the gesture sequence as a one-handed sequence; and execute a command in response to the one-handed sequence.
The present invention(s) is/are illustrated by way of example and is/are not limited by the accompanying figures. Elements in the figures are illustrated for simplicity and clarity, and have not necessarily been drawn to scale.
Embodiments described herein provide systems and methods for enabling one-handed gesture sequences in virtual, augmented, and mixed reality (xR) applications. These techniques are particularly useful in xR applications that employ HMDs, Heads-Up Displays (HUDs), and eyeglasses—collectively referred to as “HMDs.”
As used herein, the term SLAM refers systems and methods that use positional tracking devices to construct a map of an unknown environment where an HMD is located, and that simultaneously identifies where the HMD is located, its orientation, and/or pose.
Generally, SLAM methods implemented in connection with xR applications may include a propagation component, a feature extraction component, a mapping component, and an update component. The propagation component may receive angular velocity and accelerometer data from an Inertial Measurement Unit (IMU) built into the HMD, for example, and it may use that data to produce a new HMD position and/or pose estimation. A camera (e.g., a depth-sensing camera) may provide video frames to the feature extraction component, which extracts useful image features (e.g., using thresholding, blob extraction, template matching, etc.), and generates a descriptor for each feature. These features, also referred to as “landmarks,” are then fed to the mapping component.
The mapping component may be configured to create and extend a map, as the HMD moves in space. Landmarks may also be sent to the update component, which updates the map with the newly detected feature points and corrects errors introduced by the propagation component. Moreover, the update component may compare the features to the existing map such that, if the detected features already exist in the map, the HMD's current position may be determined from known map points.
In environment 100, the xR application may include a subset of components or objects operated by HMD 102 and another subset of components or objects operated by host IHS 103. Particularly, host IHS 103 may be used to generate digital images to be displayed by HMD 102. HMD 102 transmits information to host IHS 103 regarding the state of user 101, such as physical position, pose or head orientation, gaze focus, etc., which in turn enables host IHS 103 to determine which image or frame to display to the user next, and from which perspective.
As user 101 moves about environment 100, changes in: (i) physical location (e.g., Euclidian or Cartesian coordinates x, y, and z) or translation; and/or (ii) orientation (e.g., pitch, yaw, and roll) or rotation, cause host IHS 103 to effect a corresponding change in the picture or symbols displayed to user 101 via HMD 102, in the form of one or more rendered video frames.
Movement of the user's head and gaze may be detected by HMD 102 and processed by host IHS 103, for example, to render video frames that maintain visual congruence with the outside world and/or to allow user 101 to look around a consistent virtual reality environment. In some cases, xR application components executed by HMD 102 and IHSs 103 may provide a cooperative, at least partially shared, xR environment between a plurality of users. For example, each user may wear their own HMD tethered to a different host IHS, such as in the form of a video game or a productivity application (e.g., a virtual meeting).
To enable positional tracking for SLAM purposes, HMD 102 may use wireless, inertial, acoustic, or optical sensors. And, in many embodiments, each different SLAM method may use a different positional tracking source or device. For example, wireless tracking may use a set of anchors or lighthouses 107A-B that are placed around the perimeter of environment 100 and/or one or more tokens 106 or tags 110 that are tracked; such that HMD 102 triangulates its position and/or state using those elements. Inertial tracking may use data from accelerometers and gyroscopes within HMD 102 to find a velocity (e.g., m/s) and position of HMD 102 relative to some initial point. Acoustic tracking may use ultrasonic sensors to determine the position of HMD 102 by measuring time-of-arrival and/or phase coherence of transmitted and receive sound waves.
Optical tracking may include any suitable computer vision algorithm and tracking device, such as a camera of visible, infrared (IR), or near-IR (NIR) range, a stereo camera, and/or a depth camera. With inside-out tracking using markers, for example, camera 108 may be embedded in HMD 102, and infrared markers 107A-B or tag 110 may be placed in known stationary locations. With outside-in tracking, camera 105 may be placed in a stationary location and infrared markers 106 may be placed on HMD 102 or held by user 101. In others cases, markerless inside-out tracking may use continuous searches and feature extraction techniques from video frames obtained by camera 108 (e.g., using visual odometry) to find natural visual landmarks (e.g., window 109) in environment 100.
In various embodiments, data obtained from a positional tracking system and technique employed by HMD 102 may be received by host IHS 103, which in turn executes the SLAM method of an xR application. In the case of an inside-out SLAM method, for example, an xR application receives the position and orientation information from HMD 102, determines the position of selected features in the images captured by camera 108, and corrects the localization of landmarks in space using comparisons and predictions.
An estimator, such as an Extended Kalman filter (EKF) or the like, may be used for handling the propagation component of an inside-out SLAM method. A map may be generated as a vector stacking sensors and landmarks states, modeled by a Gaussian variable. The map may be maintained using predictions (e.g., when HMD 102 moves) and/or corrections (e.g., camera 108 observes landmarks in the environment that have been previously mapped). In other cases, a map of environment 100 may be obtained, at least in part, from cloud 104.
As shown, HMD 102 includes processor 201. In various embodiments, HMD 102 may be a single-processor system, or a multi-processor system including two or more processors. Processor 201 may include any processor capable of executing program instructions, such as a PENTIUM series processor, or any general-purpose or embedded processors implementing any of a variety of Instruction Set Architectures (ISAs), such as an x86 ISA or a Reduced Instruction Set Computer (RISC) ISA (e.g., POWERPC, ARM, SPARC, MIPS, etc.).
HMD 102 includes chipset 202 coupled to processor 201. In certain embodiments, chipset 202 may utilize a QuickPath Interconnect (QPI) bus to communicate with processor 201. In various embodiments, chipset 202 provides processor 201 with access to a number of resources. For example, chipset 202 may be coupled to network interface 205 to enable communications via various wired and/or wireless networks.
Chipset 202 may also be coupled to display controller or graphics processor (GPU) 204 via a graphics bus, such as an Accelerated Graphics Port (AGP) or Peripheral Component Interconnect Express (PCIe) bus. As shown, graphics processor 204 provides video or display signals to display 206.
Chipset 202 further provides processor 201 and/or GPU 204 with access to memory 203. In various embodiments, memory 203 may be implemented using any suitable memory technology, such as static RAM (SRAM), dynamic RAM (DRAM) or magnetic disks, or any nonvolatile/Flash-type memory, such as a solid-state drive (SSD) or the like. Memory 203 may store program instructions that, upon execution by processor 201 and/or GPU 204, present an xR application to user 101 wearing HMD 102.
Other resources coupled to processor 201 through chipset 202 may include, but are not limited to: positional tracking system 210, gesture tracking system 211, gaze tracking system 212, and inertial measurement unit (IMU) system 213.
Positional tracking system 210 may include one or more optical sensors (e.g., a camera 108) configured to determine how HMD 102 moves in relation to environment 100. For example, an inside-out tracking system 210 may be configured to implement markerless tracking techniques that use distinctive visual characteristics of the physical environment to identify specific images or shapes which are then usable to calculate HMD 102's position and orientation.
Gesture tracking system 211 may include one or more cameras or optical sensors that enable user 101 to use their hands for interaction with objects rendered by HMD 102. For example, gesture tracking system 211 may be configured to implement hand tracking and gesture recognition in a 3D-space using a gesture camera mounted on HMD 102, such as camera 108. In some cases, gesture tracking system 211 may track a selectable number of degrees-of-freedom (DOF) of motion, with depth information, to recognize gestures (e.g., swipes, clicking, tapping, grab and release, etc.) usable to control or otherwise interact with xR applications executed by HMD 102, and various one and two-handed gesture sequences described in more detail below.
Additionally, or alternatively, gesture tracking system 211 may include one or more ultrasonic sensors 111 configured to enable Doppler shift estimations of a reflected acoustic signal's spectral components.
Gaze tracking system 212 may include an inward-facing projector configured to create a pattern of infrared or (near-infrared) light on the user's eyes, and an inward-facing camera configured to take high-frame-rate images of the eyes and their reflection patterns; which are then used to calculate the user's eye's position and gaze point. In some cases, gaze detection or tracking system 212 may be configured to identify a direction, extent, and/or speed of movement of the user's eyes in real-time, during execution of an xR application.
IMU system 213 may include one or more accelerometers and gyroscopes configured to measure and report a specific force and/or angular rate of the user's head. In some cases, IMU system 212 may be configured to a detect a direction, extent, and/or speed of rotation (e.g., an angular speed) of the user's head in real-time, during execution of an xR application.
Transmit (Tx) and receive (Rx) transducers and/or transceivers 214 may include any number of sensors and components configured to send and receive communications using different physical transport mechanisms. For example, Tx/Rx transceivers 214 may include electromagnetic (e.g., radio-frequency, infrared, etc.) and acoustic (e.g., ultrasonic) transport mechanisms configured to send and receive communications, to and from other HMDs, under control of processor 201. Across different instances of HMDs, components of Tx/Rx transceivers 214 may also vary in number and type of sensors used. These sensors may be mounted on the external portion of frame of HMD 102, for example as sensor 111, to facilitate direct communications with other HMDs.
In some implementations, HMD 102 may communicate with other HMDs and/or host IHS 103 via wired or wireless connections (e.g., WiGig, WiFi, etc.). For example, if host IHS 103 has more processing power and/or better battery life than HMD 102, host IHS 103 may be used to offload some of the processing involved in the creation of the xR experience.
For purposes of this disclosure, an IHS may include any instrumentality or aggregate of instrumentalities operable to compute, calculate, determine, classify, process, transmit, receive, retrieve, originate, switch, store, display, communicate, manifest, detect, record, reproduce, handle, or utilize any form of information, intelligence, or data for business, scientific, control, or other purposes. For example, an IHS may be a personal computer (e.g., desktop or laptop), tablet computer, mobile device (e.g., Personal Digital Assistant (PDA) or smart phone), server (e.g., blade server or rack server), a network storage device, or any other suitable device and may vary in size, shape, performance, functionality, and price. An IHS may include Random Access Memory (RAM), one or more processing resources such as a Central Processing Unit (CPU) or hardware or software control logic, Read-Only Memory (ROM), and/or other types of nonvolatile memory. Additional components of an IHS may include one or more disk drives, one or more network ports for communicating with external devices as well as various I/O devices, such as a keyboard, a mouse, touchscreen, and/or a video display. An IHS may also include one or more buses operable to transmit communications between the various hardware components.
In various embodiments, HMD 102 and/or host IHS 103 may not include each of the components shown in
Gesture Sequence Recognition
HMDs are starting to find widespread use in the workplace, enabling new user modalities in fields such as construction, design, and engineering; as well as in real-time critical functions, such as first-responders and the like. In these types of applications, it may be desirable to perform various application or User Interface (UI) actions, through the use of gesture sequences, for expediting purposes and/or for convenience. While menus (e.g., a list of options or commands presented to the user) and voice commands (less feasible in noisy environments) are still useful, gesture sequences are valuable expediting input mechanisms.
Given the wide range of use-cases, it becomes desirable for the HMD wearer to be able to execute one-handed and two-handed gesture sequences, depending upon context, operation, and application. It therefore becomes critical to be able to recognize, differentiate between one-handed and two-handed gesture modes, and to be able to track through the start, motion, and end phases of a gesture sequence. The recognized gesture sequence may be mapped to an action (e.g., minimize workspace), to an application (e.g., a video game or productivity software), or to an UI command (e.g., menu open).
Generally, gesture detection begins when video frame data (e.g., a video or depth-video stream) is received at host IHS 103 from camera 108 of HMD 102. In some implementations, the video data may have already been processed, to some degree, by gesture tracking component 211 of HMD 102. Then, the video data is further processed by gesture sequence recognition component 301 using calibration component 302 to control aspects of xR application 303, by identifying various gestures and gesture sequences that constitute user input to xR application 303, as further described below.
Generally, at least a portion of user 101 may be identified in the video frame data obtained using camera 108 using gesture sequence recognition component 301. For example, through image processing, a given locus of a video frame or depth map may be recognized as belonging to user 101. Pixels that belong to user 101 (e.g., arms, hands, fingers, etc.) may be identified, for example, by sectioning off a portion of the video frame or depth map that exhibits above-threshold motion over a suitable time scale, and attempting to fit that section to a generalized geometric model of user 101. If a suitable fit is achieved, then pixels in that section may be recognized as those of user 101.
In some embodiments, gesture sequence recognition component 301 may be configured to analyze pixels of a video frame or depth map that correspond to user 101, in order to determine what part of the user's body each pixel represents. A number of different body-part assignment techniques may be used. In an example, each pixel of the video frame or depth map may be assigned a body-part index. The body-part index may include a discrete identifier, confidence value, and/or body-part probability distribution indicating the body part or parts to which that pixel is likely to correspond.
For example, machine-learning may be used to assign each pixel a body-part index and/or body-part probability distribution. Such a machine-learning method may analyze a user with reference to information learned from a previously trained collection of known gestures and/or poses stored in calibration component 302. During a supervised training phase, for example, a variety of gesture sequences may be observed, and trainers may provide label various classifiers in the observed data. The observed data and annotations may then be used to generate one or more machine-learned algorithms that map inputs (e.g., observation data from a depth camera) to desired outputs (e.g., body-part indices for relevant pixels).
Thereafter, a partial virtual skeleton may be fit to at least one body part identified. In some embodiments, a partial virtual skeleton may be fit to the pixels of video frame or depth data that correspond to a human arm, hand, and/or finger(s). A body-part designation may be assigned to each skeletal segment and/or each joint. Such virtual skeleton may include any type and number of skeletal segments and joints, including each individual finger).
In some embodiments, each joint may be assigned a number of parameters, such as, for example, Cartesian coordinates specifying joint position, angles specifying joint rotation, and other parameters specifying a conformation of the corresponding body part (e.g., hand open, hand closed, etc.), etc. Skeletal-fitting algorithms may use the depth data in combination with other information, such as color-image data and/or kinetic data indicating how one locus of pixels moves with respect to another. Moreover, a virtual skeleton may be fit to each of a sequence of frames of depth video. By analyzing positional change in the various skeletal joints and/or segments, certain corresponding movements that indicate predetermined gestures, actions, or behavior patterns of user 101 may be identified.
In other embodiments, the use of a virtual skeleton may not be necessary. For example, in other implementations, raw point-cloud data may be sent directly to a feature extraction routine within gesture sequence recognition component 301.
Particularly, each “gesture sequence,” as the term is used herein, has a Start phase (S) with a standalone gesture, a motion phase (M) with a sequence of gestures following each other, and an end phase (E) with another standalone gesture. In some cases, E may be the last gesture in the sequence (end state) of M.
In some embodiments, a look-up tables may be used to store key attributes and/or reference images of start, motion, and end phases for each gesture sequence to be recognized, for two-handed and one-handed cases. As used herein, the term “look-up table” or “LUT” refers to an array or matrix of data that contains items that are searched. In many cases, LUTs may be arranged as key-value pairs, where the keys are the data items being searched (looked up) and the values are either the actual data or pointers to where the data are located.
A setup or calibration phase may store user-specific finger/hand/arm attributes (e.g., asking user 101 to splay fingers), such as motion velocity or asynchrony. For example, a start or end phase LUT may include reference images or attributes, whereas a motion phase LUT may include relative 6-axes or the like.
The amount of time user 101 has to hold their hands and/or fingers in position for each phase of gesture sequence (S, M, and E) may be configurable: the number of gesture camera frames for each phase may be given by F_S, F_M, and F_E, respectively. In many cases, F_S and F_E should be greater 1 second, wherein F_M should be at least 1 second (e.g., at 30 fps, a configuration may be F_S=30, F_M=45, F_E=30).
At block 401, method 400 is in sleep mode, and it may be woken in response to infrared IR or ALS input. At block 402, method 400 starts in response to IR/ALS detection of a decrease in captured light intensity (e.g., caused by the presence of a hand of user 101 being placed in front of HMD 102). At block 403, method 400 recognizes a one or two-handed gesture sequence start. At block 404, method 400 performs recognition of the start phase of a gesture sequence (or it times out to block 401). At block 405, method 400 performs recognition of the motion phase of the gesture sequence (or it times out to block 401). And, at block 406, method 400 performs recognition of the end phase of the gesture sequence (or it times out to block 401).
Then, at block 407, method 400 identifies the gesture sequence and initiates a corresponding action. In some embodiments, method 400 may be used to launch a process, change a setting of the host IHS 103′s Operating System (OS), shift input focus from one process to another, or facilitate other control operations with respect to xR application 303. The mapping of each gesture-hand-S-M-E tuple to an action may be performed by a user, or it may be set by default.
As such, method 400 receives gesture camera frame inputs, recognizes states (S, M, E) of a gesture sequence, differentiates between one-handed and two-handed sequences on each frame, tracks the state of start, motion, and end phases using hysteresis and LUTs with timeouts, and maps a recognized gesture to an action in UI or application.
Particularly, method 500 begins at block 501 in sleep mode (e.g., waiting for an ALS event, or the like). At block 502, in response to the ALS detecting a change of light within a region-of-interest (ROI) in front of HMD 102, method 500 starts a histogram of [gesture detected, handedness detected] across a maximum number of F_S, F_M, and F_E frames as a buffer size allows.
At block 503, method 500 may extract a first set of features from each of a first set of video frames and it may count the number of fingers on each hand, for example, using segmentation techniques to detect different hands and to recognize a gesture instance. For example, a hand may be first detected using a background subtraction method and the result of hand detection may be transformed to a binary image. Fingers and palm may be segmented to facilitate detection and recognition. And hand gestures may be recognized using a rule classifier.
If block 504 determines that neither a one nor a two-handed gesture has been recognized, control returns to block 501. For example, an ALS event may be determined to be a false alarm if user 101 accidentally moves his or her hand (or moves it without intention to gesture) or if the ambient light actually changes.
Conversely, if block 504 determines that a gesture has been recognized, this marks the possible beginning of the start phase of the gesture sequence. At block 505, method 500 collects [gesture detected, handed-mode detected] tuple over F_S frames. Then, at block 506, method 500 runs a histogram to find a most frequently detected gesture G1, and a second most detected gesture G2. At block 507, if G1 is not greater than G2 by at least a selected amount (e.g., 3 times to ensure accuracy), control returns to block 501.
If block 507 determines that G1 is greater than G2 by the selected amount, this confirms the start phase and begins the motion phase of the gesture sequence. At block 508, method 500 analyzes motion, movement, and/or rotation (e.g., using 6-axes data) collected over F_M frames. For example, block 508 may apply a Convolutional Neural Network (CNN) that handles object rotation explicitly to jointly perform object detection and rotation estimation operations.
At block 509, method 500 looks up 6-axes data over F_M frames against calibration data for the motion phase. A calibration procedure may address issues such each user having different sized fingers and hands, and also how different users execute the same motion phase with different relative velocities. Accordingly, a user-specific calibration procedure may be performed by calibration component 302 prior to execution of method 500, may measure and store user's typical velocity or speed in motion phase across different gestures.
At block 510, if user 101 has not successfully continued the gesture sequence's start phase into the motion phase with correct 6-axes relative motion of arm(s), hand(s), and/or finger(s), control returns to block 501. Otherwise, block 510 confirms the motion phase and begins the end phase of the gesture sequence. At block 511, method 500 may extract a second set of features from each of a second set of video frames to count the number of fingers on each hand, for example, using segmentation techniques to detect different hands and to recognize a gesture instance. If block 512 determines that the recognized gesture is not the end state of the gesture sequence being tracked, control returns to block 501.
Otherwise, at block 513, method 500 collects [gesture detected, handed-mode detected] tuple over F_E frames. At block 514, method 500 runs a histogram to find a most frequently detected gesture G1, and a second most detected gesture G2. At block 515, if G1 is not greater than G2 by at least a selected amount (e.g., 3 times to ensure accuracy), control returns to block 501. Otherwise, at block 516, the end phase of the gesture sequence is confirmed and the sequence is successfully identified, which can result, for example, in an Application Programming Interface (API) call to application or UI action at block 517.
Accordingly, in various embodiments, method 500 may include receiving a gesture sequence from an HMD configured to display an xR application, and identifying the gesture sequence as: (i) a one-handed gesture sequence, or (ii) a two-handed gesture sequence.
In an example use-case, user 101 may start with a one-handed gesture sequence, but then decides to perform a different gesture sequence with the other hand. In that case, method 500 may be capable of detecting two distinct one-handed gesture sequences applied in serial order, instead of a single two-handed gesture sequence.
Particularly, LUTs may be used to detect differences between a two-handed gesture-sequence start versus two one-handed gesture-sequence start. Method 500 may use multiple frames of logical “AND” and “MAX” operations along with user-specific calibration detect two hands in at least one of the second set of video frames; and in response to a comparison between the first and second sets of features, method 500 may identify the gesture sequence as a first one-handed gesture sequence made with one hand followed by a second one-handed gesture sequence made with another hand. For gesture sequences that generate false positives, a serial non-reentrant variant of method 500 may be implemented.
In another example use-case, user 101 may perform a two-handed gesture sequence, but due to natural human asynchronicity (e.g., both hands do not start simultaneously), user 101 may have a time difference between the hands gesture being detected by or appearing in front of HMD 102. In those cases, method 500 may detect only one hand in at least one of the second set of video frames; and in response to a comparison between the first and second sets of features, method 500 may identify the gesture sequence as a two-handed gesture sequence. To achieve this, method 500 may perform a table look-up operation using calibration data that includes an indication of the user's asynchronicity (e.g., an amount of time in milliseconds) during the motion phase.
In sum, method 500 allows users to use gesture sequences with one or two hands for different purposes, where voice commands or menu are not feasible to use (e.g., noisy environments, like factories) or where silence is essential without giving away location (e.g., first responders). In various embodiments, method 500 may be implemented without hardware dependency or additions, by using host-based algorithms, as a cross-platform service or application, with API access for recognized gestures.
At block 603, user 101 performs a gesture sequence to enable a selected action, records the received audio signal, de-noises it, and filters the discrete frequencies. Block 604 stores the amplitude decrease (in the frequency domain) for both of the user's ears, with transceivers 111 on both sides of HMD 102. At block 605, user 101 performs another gesture sequence to disable the selected action, records another received audio signal, de-noises it, and again filters the discrete frequencies. Block 606 stores the amplitude increase (in the frequency domain) for both ears, and method 600 may return to block 603 for the calibration of another gesture sequence.
At block 801, method 800 buffers a received audio pattern (resulting from the transmission of block 702), de-noises it, and filters by the three frequencies across sliding windows of N seconds, to perform Doppler shift estimations of the measured signal spectral components. At block 802, method 800 performs pattern matching operations against other stored patterns. Then, at block 803, if the received pattern is recognized, the gesture sequence is identified, and control returns to block 801.
In various embodiments, methods 600-800 may be performed for gesture sequences that take place at least partially outside the field-of-view of a gesture camera, for example, near the side of the user's head. Moreover, visual gesture sequence recognition method 500 and ultrasonic gesture sequence recognition methods 600-800 may be combined, in a complementary manner, to provide a wider range of gesturing options to user 101.
One-Handed Gesture Sequences
Systems and methods described herein may be used to enable different types of one-handed gesture sequence recognition in xR HMDs.
In some embodiments, gesture sequence detection system 300 may employ methods 600-800 to perform ultrasonic-based (111) detection of one-handed gesture sequences. For example,
In gesture sequence 900A, starting position 901 shows a user covering her left ear 910 with the palm of her left hand 909, followed by motion 902, where the left hand 909 moves out and away from the user, uncovering her left ear 910. In response to detecting gesture sequence 900A, a previously muted left audio channel (e.g., reproduced by a left speaker in the HMD) is unmuted; without changes to the operation of the right audio channel. Additionally, or alternatively, a volume of the audio being reproduced in the left audio channel may be increased, also without affecting the right audio channel.
In gesture sequence 900B, starting position 903 shows a user with her left hand 909 out and away by a selected distance, followed by motion 904 where the user covers her left ear 910 with the palm of her left hand 909. In response to detecting gesture sequence 900B, the left audio channel may be muted, or its volume may be reduced, without changes to the operation of the right audio channel.
In gesture sequence 900C, starting position 905 shows a user covering her right ear 912 with the palm of her right hand, followed by motion 906 where the right hand 911 moves out and away from the user, uncovering her right ear 912. In response to detecting gesture sequence 900C, a previously muted right audio channel (e.g., reproduced by the right speaker within the HMD) is unmuted, or a volume of the audio being reproduced in the right audio channel may be increased, without affecting the left audio channel.
In gesture sequence 900D, starting position 907 shows a user with her right hand 911 out and away by a selected distance, followed by motion 908, where the user covers her right ear 912 with the palm of her right hand 911. In response to detecting gesture sequence 900D, the right audio channel may be muted, or its volume may be reduced, without changes to the operation of the left audio channel.
In various other implementations, gesture sequence detection system 300 may employ methods 400 and/or 500 to perform camera-based detection of one-handed gesture sequences.
For example,
In deselect gesture sequence 1000B, frame 1006 shows an already selected xR object 1010 (e.g., using gesture sequence 1000A) with user's hand 1011 with palm out hovering over xR object 1010 with fingers spread apart. Another start phase, illustrated in frame 1007, follows on with fingers closed together over xR object 1010. After a selected period of time, such as during a motion phase, visual effect 1013 (e.g., highlighting) may be removed from xR object 1010 (e.g., surround highlighting may flash and then fade away), as shown in frame 1008, to indicate that xR object 1010 has been deselected. End frame 1009 shows xR object 1010 remaining on the HMD's display, still rotated, after the user's hand disappears from sight, at which time object menu 1014 may also be removed.
If hand 1220 moves away before highlighting 1222 disappears, menu 1221 is dismissed. If hand 1220 stays in position until highlighting 1222 disappears (after a timeout), as shown in frame 1203, subsequent movement of the left hand 1220 does not affect the display of menu 1221, as shown in frame 1204 (still available for voice or hand selection). Conversely, menu open gesture sequence 1200B of
Close menu gesture sequence 1200C shows, as its initial condition, open menu 1221 in frame 1209. Start phase with frame 1210 shows a user's right hand 1223 held up with palm facing in with all five fingers held together in a cusped position behind the open menu 1221. After a selected period of time, such as during a motion phase, the contents of open menu 1221 are dimmed, and surround highlighting 1222 appears, as shown in frame 1211. Then, at the end phase shown in frame 1212, the fingers of hand 1223 close, in a grabbing motion, and menu 1221 is also closed.
Menu repositioning gesture sequence 1200D begins with initial frame 1213 showing an open menu 1221 on the HMD display. Start phase with frame 1214 shows a user's right hand 1223 held up with palm facing in with all five fingers held together in a cusped position behind open menu 1221. After a selected period of time, such as during a motion phase, the contents of the open menu are dimmed, and surround highlighting 1222 appears as shown in frame 1215. Still during the motion phase, as shown in frame 1216, open menu 1221 follows or tracks the user's hand, on the HMD display, as the user's hand 1223 correspondingly moves across the front of the HMD display. When movement of hand 1223 stops for a predetermined amount of time, highlighting 1222 disappears. At the new position in frame 1217, the user removes her hand from sight, and menu 1221 is maintained at its new position.
Additionally, or alternatively, an xR application may provide a workspace that enables interactions of VO (or groups of VOs) in a workspace with real world and viewing VOs, including, but not limited to: hiding/showing layer(s) of VOs, foreground/background occlusion (notion of transparency and ability to highlight interference for VO on VO or VOs on physical object), texture and materials of VOs, cloning a real world object into VO and printing a VO to model of real world prototype, identification of object as VO or real, physics treatments around workspaces involving VOs including display level manipulations (e.g., show raindrops on display when simulating viewing a car VO in rain) etc.
Additionally, or alternatively, an xR application may provide a workspace that enables operations for single VO or group of VOs in workspace, including, but not limited to: rotate, resize, select, deselect (VO or layers within), lock/unlock a VO for edits or security/permissions reasons, grouping of VOs or ungrouping, VO morphing (ability to intelligently resize to recognize dependencies), copy/paste/delete/undo/redo, etc.
Additionally, or alternatively, an xR application may enable operations on entire workspaces, including, but not limited to: minimize/maximize a workspace, minimize all workspaces, minimize/hide a VO within a workspace, rename workspace, change order of workspace placeholders, copy VO from one workspace to another or copy an entire workspace to another, create a blank workspace, etc. As used, herein, the term “minimize” or “minimizing” refers to the act of removing a window, object, application, or workspace from a main display area, collapsing it into an icon, caption, or placeholder. Conversely, the term “maximize” or “maximizing” refers to the act of displaying or expanding a window, object, application, or workspace to fill a main display area, for example, in response to user's selection of a corresponding icon, caption, or placeholder.
In some cases, an xR application may enable communication and collaboration of workspaces, including, but not limited to: saving a workspace to a file when it involves multiple operations, multiple VO, etc.; being able to do differences across workspaces (as a cumulative) versions of same file, optimized streaming/viewing of workspace for purposes of collaboration and live feedback/editing across local and remote participants, annotating/commenting on workspaces, tagging assets such as location etc. to a workspace, etc.; includes user privileges, such as read/write privileges (full access), view-only privileges (limits operations and ability to open all layers/edits/details), annotation access, etc.
Back to
First down/up minimization gesture sequence 1300C of
Down/up maximization gesture sequence 1300E of
Initial condition 1401 shows xR object 1410 on the HMD display. The start phase of frame 1402 shows left hand 1411 making an index selection of a portion of the xR object 1412 (e.g., a wheel), which is shown highlighted. At frame 1403, during a motion phase, left hand 1411 performs an annotation gesture with an index finger pointing out, in the middle of the display, while the wheel 1412 is still highlighted.
In response to the annotation gesture, frame 1404 shows a text bubble and blinking cursor 1413 that allows the user to enter data 1414 associated with selected object 1412 (e.g., by speaking and having the speech translated into text using a voice service), as shown in frame 1405. Frame 1406 illustrates the use of comment icons 1414 for previously entered annotations. Selecting an annotation, as shown in frame 1407 (e.g., touching with fingertip), causes a text and associated xR region 1414 to be displayed, as in frame 1406. In some cases, holding a fingertip on a comment icon for a time duration (e.g., 3 seconds), until the icon flashes and highlights, allows the user to reposition the icon on the display.
For example, in lock gesture sequence 1700A, start frame 1701 illustrates hand 1705 hovering over xR object 1706 to be locked, with its index and middle fingers in their extended positions, and the other three fingers curled or retracted. During a motion phase, as shown in frame 1702, the user crosses the index and middle fingers over xR object 1706. In response, access to xR object 1706 is locked. Conversely, in unlock gesture sequence 1700B, start frame 1703 shows hand 1705 hovering over xR object 1706 to be unlocked with index and middle fingers crossed, followed by a motion phase where the index and middle fingers assume an extended position, as in frame 1704. In response, access to xR object 1706 is then unlocked.
In multi-user collaboration of workspaces, these gesture sequences may be used by any collaborator to lock/unlock access temporarily for purposes of illustration, discussion, editing, as long as the owner has started the collaboration mode with a two-handed gesture sequence. When this gesture is used by an owner to lock, control may be immediately turned over to owner from any other collaborator who has locked the item/workspace. When these gestures are used by owner to unlock, the current collaborator's lock may be undone, and the workspace becomes unlocked for all participants. Moreover, in some cases, when a collaborator has locked an item or workspace and another collaborator wants to lock same item or workspace, the prior collaborator may need to unlock it first, or the owner may have to unlock pre-emptively.
Locked items may include, but are not limited to, workspaces (e.g., file selection shown), virtual or digitally-generated xR objects, regions of xR objects (e.g., a wheel of a car), or an edit layer in an object menu. When a collaborator has locked item or workspace, the owner may unlock it without any permissions pre-emptively. In sequence 1800B, a layer or filter may be turned on in frame 1803 to provide a visual indicator of locked or disabled items 1809 (e.g., grayed-out items) in frame 1804.
Two-Handed Gesture Sequences
Systems and methods described herein may be used to enable two-handed gesture sequence recognition in xR HMDs.
In various embodiments, gesture sequence detection system 300 may employ methods 400 and/or 500 to perform camera-based (108) detection of two-handed gesture sequences. For example,
In “display off” gesture sequence 2000B, start frame 2003 illustrates a user with two hands 2005 and 2006 in the same position as in frame 2002. During a subsequent motion phase, as shown in frame 2004, the user covers his or her face 2008, as previously shown in frame 2001. In response, the user's HMD display may be turned off, or the HMD may be placed in a sleep or low power state. Frame 2000C shows a side view 2007 of the user, and a typical distance between the user's hands and the HMD during gesture sequences 2000A and 2000B.
In gesture sequence 2200B, start frame 2204 shows hands 2207 and 2208 making the same frame shape, but around an existing display overlay region 2211 that is no longer wanted. After a predetermined amount of time, such as during a motion phase shown in frame 2205, the display overlay region 2211 actively reproducing a video is dimmed and the region is highlighted with visual effect 2212. At end phase 2206, the user separates hand 2207 from hand 2208, and display overlay region 2211 is dismissed.
Conversely, open gesture sequence 2400B of
Particularly, in gesture sequence 2600A, start phase frame 2601 illustrates first cusped hand 2612 with palm facing up, and second cusped hand 2613 with palm facing up and joining the first cusped hand at an obtuse angle. After holding still for a predetermined amount of time, such as during a motion phase, frame 2602 shows virtual key 2614 (or other xR object) rendered over the user's hands 2612 and 2613.
To a collaborator wearing a different HMD, another virtual key 2615 appears on their display in frame 2603. During a motion phase performed by the collaborator, as shown in frame 2604, the collaborator makes a grabbing gesture 2616 to take virtual key 2615 and become the new “active” user with the ability to manipulate or edit the workspace.
After another predetermined amount of time, virtual key 2617 may serve as an indicator to other collaborators, and it may be shown in other participants' HMDs over the active user's head, for example, as shown in frame 2605 of sequence 2600B in
In a session with multiple participants, each participant wearing their own HMD, gesture sequence 2600C starts with frames 2606 and 2607, akin to frames 2601 and 2602 in
In some implementations, gesture sequence detection system 300 may employ methods 600-800 to perform ultrasonic-based detection of two-handed gesture sequences.
It should be understood that various operations described herein may be implemented in software executed by logic or processing circuitry, hardware, or a combination thereof. The order in which each operation of a given method is performed may be changed, and various operations may be added, reordered, combined, omitted, modified, etc. It is intended that the invention(s) described herein embrace all such modifications and changes and, accordingly, the above description should be regarded in an illustrative rather than a restrictive sense.
Although the invention(s) is/are described herein with reference to specific embodiments, various modifications and changes can be made without departing from the scope of the present invention(s), as set forth in the claims below. Accordingly, the specification and figures are to be regarded in an illustrative rather than a restrictive sense, and all such modifications are intended to be included within the scope of the present invention(s). Any benefits, advantages, or solutions to problems that are described herein with regard to specific embodiments are not intended to be construed as a critical, required, or essential feature or element of any or all the claims.
Unless stated otherwise, terms such as “first” and “second” are used to arbitrarily distinguish between the elements such terms describe. Thus, these terms are not necessarily intended to indicate temporal or other prioritization of such elements. The terms “coupled” or “operably coupled” are defined as connected, although not necessarily directly, and not necessarily mechanically. The terms “a” and “an” are defined as one or more unless stated otherwise. The terms “comprise” (and any form of comprise, such as “comprises” and “comprising”), “have” (and any form of have, such as “has” and “having”), “include” (and any form of include, such as “includes” and “including”) and “contain” (and any form of contain, such as “contains” and “containing”) are open-ended linking verbs. As a result, a system, device, or apparatus that “comprises,” “has,” “includes” or “contains” one or more elements possesses those one or more elements but is not limited to possessing only those one or more elements. Similarly, a method or process that “comprises,” “has,” “includes” or “contains” one or more operations possesses those one or more operations but is not limited to possessing only those one or more operations.
