According to the writer Wang Rui, VR senior industry practitioners.
Existing construction of VR experience has left out the most important part, is cheap and flexible and accurate positioning solutions. Described in this paper have some advantages and disadvantages of the. This neck piece.
Storm has intensified inside virtual reality, from a wide variety of helmets and glasses, to a variety of strange ideas of interactive devices, content production and attempts to build experience. An increasing number of VR related practitioners and entrepreneurs will be The Void,Zero Latency such as their next goal of overtaking, VR theme park development was issued, or VR theme experience solution’s declaration.
However, from macro of angle view, whole industry also is not development of smooth: VR helmet and glasses of standard uneven, mainstream products (for example OculusVR and HTC Vive) yet completely expand consumers market; VR on system configuration of requirements is high, related equipment carry wear easily, and mobile end equipment of rendering is most force has not caught; and for experience Museum of interactive content how making, interactive way how design, also has many not is clear of place. In addition, construction of VR experience has left out the most important part, is cheap and flexible and accurate positioning solutions.
Here said the position, is to determine the experience of participants in absolute spatial location in the venue, and feedback to all players involved in the game and the game server, game logic necessary to perform community games.
For example, when the player is near the edge of the Woods, a deliberate hungry wolves would suddenly rush out; for example a player more live-action CS of virtual reality contest, shooting and tactics. If the player could not identify their location in the playing field, so naturally a lot less fun and complexity, may just be another first-person fixed shooting game. Positioning accuracy and speed cannot be overlooked, 20cm whether the error may have decided that a bullet can penetrate the virtual enemy’s chests, positioning itself delayed, will also give the player’s feeling of virtual reality have a huge impact, even become accomplices of 3D motion.
However, throughout VR space orientation of various programmes on the market now, but none provide enough mature and stable technology, or with huge costs and the cost of the loss of flexibility, build is not successful Demo works. Perhaps far more so cannot be a solution, and nearly enough to satisfy the appetite of many experience practitioners. So, good positioning solution should be your goals to what standard? The industry as a whole is the missing Savior has come? This article will try to slightly in detail.
(A) the body cameras
Late last year, Japan Shinjuku bustling conducted a one-day VR experience, by wearing Gear VR Displays a virtual snow scenes and props of the built entities suspension and simulation through the fan and air conditioning cold mountain environments. Experience gingerly forward on the bridge itself, and final position at the end of the field to get their prizes–a hot drink from the dealer.
The complexity of the game is not high, its effects are very significant. An indispensable link and one is judging players walk on the bridge position, and Gear corresponding to rendered scenes in VR. In the same video, we can easily find any clue:
It is clear that this is a Microsoft Kinect2 device, it passes a TOF (time of flight) method of calculation, get sensor encounter objects reflection phase after the emergent Ray to get distance between the device and the actual scene, or depth (depth). Depth of the resulting images can be used to determine relative sense of the location of an object, you can further the point cloud data of the object, even the bones of the role of information. More importantly, Kinect2 is only optional peripherals for the XBox One games, so you want to buy it and simple spatial orientation for VR content will never spend too much cost.
However, inexpensive is the opposite of adequacy of performance parameters, 30fps refresh rate so that people can clearly feel the delay of results (although this delay for delayed compared to the helmet, giving less), as well as Kinect Visual angle of about 60 degrees, maximum recognition range in General: 3-4 m. In this region can recognize up to 6 location information, and within their field of view in the Kinect can’t have too much overlap to avoid missing (as shown in above figure). Clearly, these harsh conditions made it hard for us to imagine a more complex game than climb snow-capped mountains to drink coffee, but at least this is a good place to start.
(B) the optical positioning and image recognition
With the last news, in the ears of most VR practitioners should be familiar from Australia’s Zero Latency becomes the world’s first virtual reality gaming centers, it covers an area of around 400 square metres, made up of 129 PS Eye camera and support 6 players with the game … …
Yes, the experience compared to previous VR applications, its biggest selling point is that you can walk freely in virtual space. PS Move equipment (including the PS Eye camera, Move the handle of the ball and is equipped with a tag light) is the heart of this technology.
As shown in the renderings of the game before, different colors of light in the PS Eye camera can exhibit significantly different from the background image of the image, so as to facilitate computer vision (CV) method to extract it. Of course, there is only a PS Eye camera, was unable to get the player’s three dimensional information. Need to have more than one camera to capture the player’s ball at the position screen space, and then by aerial triangulation algorithm, made true location of players in the world coordinate system.
Here nature brings out two big problems:
, How to smoothly and accurately from the camera screen in the judgments and distinguish between the different spheres of light (marker);
Second, how do I know if the camera itself in world space position and attitude, so as to correctly calculate the position and attitude of the players.
In order to solve the first problem, enough to so many developers struggle, wave upon wave. Differentiated by color point scheme is certainly feasible, but if there is a similar camera color interference? Or the site is a colorful environment do? This time, mistaken identification is probably difficult to avoid. Therefore, a number of optical motion capture vendors stood out, they choose to use an infrared camera as an alternative to recognition.
Here, of course, first established the OptiTrack, they used a frame rate greater than 100Hz of professional camera, and a global shutter system, thus avoiding the high speed moving objects on the screen, the phenomenon of motion blur. Around the camera using infrared LED light and made of highly reflective material to make players wear marks. Because the infrared camera itself have blocked most of the visible information, so tag is particularly evident in the picture. Unless there is any interference with the infrared light source to, or miscarriage of Justice off the almost impossible situation.
According to the information currently known, including The Void theme park, as well as Project Alice Connaught room, are based on the spatial orientation of the OptiTrack programme, this will undoubtedly prove the reliability of such programmes. In contrast, tends to be high cost (pictured below, a camera’s price is calculated in tens of thousands of Yuan, while building a rule-space requires at least 4 this kind of camera and software system).
However, since the luminous marking used alternative PS Move light programmes (another benefit of this is, do not consider the question of how to mark the power supply), then how to distinguish between tag ID, and distinguish between multiple players in a game? Also has a variety, for example by regulating the reflectivity, marked ball in the camera screen displays a different brightness; or a combination of different methods, a group of ball of light in the picture presents a unique combination of form, as shown in the following figure:
Said before the second question: How do I know if the camera itself in world space position and orientation. In fact is calibrated (calibration) to complete. Experience building in advance of each camera mounted to a fixed location, and then watched the footage. From a known position markers to projections of each camera position and orientation, and preserved. This process is tedious and boring, especially when you need to configure hundreds of cameras. After the Setup is complete, how to avoid the camera was moved again, or vibrate and shift in venue structural issues, this is every developers have to deal with maintenance problems.
However, because optical positioning with considerable accuracy and stability by adjusting camera parameters can also achieve very low latency, and in theory can be extended to infinite space, so it does become the current build of choice for many VR experience. However, markers to identify players or has an extremely limited, since markers must be combined, had close to two sets of markers (for example, two players fighting back), also very prone to error log or unrecognized situation. In addition, venues that are too complex environment will also make it easier for marker is clear of obstacles blocking, thus missing issues, so we can see more than the optical orientation experience, is the rule in an empty room for games.
(C) laser radar
Itself has a very precise laser ranging capability, the ranging accuracy up to mm, common is Japan HOKUYO bei Yang electric and Germany SICK Cique two-dimensional produced by professional manufacturers such as LiDAR.
Two-dimensional meaning that the light emitted by lidar, a fan-shaped plane, used for mapping purposes or construction as the three-dimensional reconstruction of three dimensional lidar, is this adds a dimension of two-dimensional laser radar spinning, to get the results of three dimensional space.
Laser radar consists of a single narrow-band laser beam and a receiver. Generates and emits a beam of laser light pulses hit objects and reflected back, finally received by the receiver. Receiver to accurately measure light pulses from launch to be reflected back to the propagation time, TOF (Time of Flight). Because optical pulses with the speed of light, so the receiver will always be received before the next pulse sent a pulse that is reflected back to before. In view of the speed of light is known, measurement of travel time can be converted to distance. Because at the speed of light for a ruler to measure, so accuracy of LIDAR are generally high, in indoor situations, errors at the millimeter level.
Two-dimensional laser radar is also led by a one-dimensional, one-beam laser on a rotating pedestal spins on the formation of fan, a two-dimensional laser radar can be self-centered, with dozens of meters for measuring RADIUS to draw fan, so if someone activities in this area, LiDAR can precisely known a person’s location and output to the computer are used. Of course, the LiDAR can also be used for research on obstacle avoidance sensors usually.
Rotating speed of the base has also been in several sizes, so laser radar can scan frequencies of distinction, normally, dozens of Hz scan frequency of position detection in VR we do enough to use.
But the principle of LIDAR high requirements on the components, usually work in very harsh conditions, which itself requires a waterproof and dustproof and tens of thousands of hours of trouble-free reliability. Production costs are not cheap. And scanning frequency, the higher the detection distance farther (that is, the transmit power, the greater) laser radar, prices are more expensive. So the price of two-dimensional laser radar has a lot of money (nearly thousands of Yuan to tens of thousands of Yuan), and Airborne Lidar mapping is not the average person can dump the (hundreds of thousands of millions).
In addition to price, laser radar for locating, there was one major problem is: because is a fan-shaped light from lasers, so if there are a lot of moving objects with each other when crowded together, will shelter each other, objects behind objects in front of the “shadow zone”, resulting in undetectable. Radar and laser ranging only, object ID is not recognized and, therefore, does not consider the price factor, also suitable for single play, if you want to group, and other programmes.
(D) the HTC Vive:Light House
In that time, HTC Vive sale became a big news in the industry as a whole. And it is a different from optical positioning is talking place. HTC Vive includes three parts, helmets and handles itself is covered with infrared sensors, as well as the positioning of the Light House. Players advance Light layout into an empty House in the two corners of the room, the two Light House equivalent of laser launch of two fixed base stations, as shown in the following figure.
Light the rear door of the House opened, like this:
The dense LED is used to synchronize the beams. Two cylinders are rotating lasers. A sweep is the x axis and the other is the y axis sweep. 180 degree phase difference of the two lasers are fixed, that is, a b, when the light is not lit, b when the light is not lit.
Handle and have a fixed location on the helmet light sensors that are installed:
The workflow of the system is divided into three steps:
1, sync: overall bright LED light Board, handles and helmet sensor being irradiated, as a synchronized signal.
2, x-axis scanning: horizontal laser irradiation of a handle and light sensors on the helmet.
3, y-axis scanning: handles Word vertical laser irradiation and photo sensors on the helmet.
Display and controller installed on a lot of light sensors. LED on the base station flashes will automatically synchronize all equipment of the time, then began laser scanning, photo sensors can measure the x axis lasers and laser reached the y axis sensor.
In other words, the laser scan across the sensor is sequenced, therefore several sensor signals on the first time there is a hierarchy between, and each sensor relative to the base station of the x axis and y axis-angle are known; and head and a handle installed on the location of the sensor has been calibrated in advance, position is fixed. Poor depending on where each sensor, you can calculate the position and trajectory of head and handle.
Light House’s biggest advantage is that it requires less computation. It is not needed as Visual CV systems imaging, then the software will feature identified in the imaging, image details, the richer the needed image processing capacity is higher. Light House using only a photosensitive device without image, does not involve a large number of computing and image processing, avoiding performance losses and the potential for instability.
The other hand, computational delay often means higher and cannot be done via embedded processor operation. The Light House because the computation and embedded systems can calculate and process, then directly to the location on the data transfer to the PC and save a lot of waste in the delay of transmission and processing. virtual reality technology
However, while VR interaction Light House is by far the best pointing device, but because the laser eye-safe laser power constraints and other problems, it can cover distances relatively limited, probably only 5M* an area of 5M square, and not have too much signal cannot be received, the curtain. And the installation of such equipment was cumbersome, and may be difficult for general users.
Not to be continued. virtual reality 3d headset