Prioridades de la Oficina de Evaluación para 2008

As mentioned in the device analysis sections, though the second prototype was a marked improvement over the first generation device, there was still scope for improvement, primarily in terms of device bulk and screen resolution, to improve user experience. HMDs catered toward virtual reality and gaming are constantly being developed by

111 today’s tech giants such as Samsung, HTC, and Oculus. Given the positioning of this device as a low-vision device based on advancements in technology, it would be important to closely monitor the VR industry for newly launched devices that could be used in the project. For this project, the idea is to buy hardware available in the general market, and develop the required software, rather than concentrating efforts on

developing the hardware. While monitoring the market for new devices, things to look out for would be:

1) Device Size and Fit: Is the device lighter than the Rift DK2? Does it offer a better form factor? Is it proven to be more comfortable?

2) All-In-One ability: Does the device have an on-board processor that eliminates the need for a laptop, making it more portable? (See GearVR + SMI device below)

3) Viewing: Does the device have a better field-of view (>100-degrees diagonal), better resolution (greater than 1920x1080) and greater number of pixels per degree? One advantage of working with low-vision subjects is that high spatial resolution is not required. For AMD patients using the peripheral retina, the critical print size is typically larger than 1 degree. Though a 10 pixel per degree resolution might prove blurry for normally sighted subjects, it would not be problematic for low vision users, so a higher field-of-view should be prioritized over a higher resolution.

4) Eye Tracking: Does HMD come with eye-trackers build in or does it have to be retro-fitted (See FOVE device below)? If the device doesn’t come with eye

112 trackers, do any eye tracking companies retro-fit this device? Three companies to contact are ArringtonResearch, EyeTechDS and SMI, as they keep abreast with newly launched HMDs. It is also important that the eye tracker images the eye along the optical axis as in the second generation device, as opposed to from an offset as in the first generation device. Next, eye-tracking frequency (in Hz) and end-to-end system latency are important. As a very basic metric, look for eye tracking frequency to be greater than the HMD’s refresh rate, and for an eye tracker processing time of just one frame (i.e. delivery of gaze coordinates to the computer is done within one frame capture by the eye tracker). Finally, binocular eye-trackers that can measure the pupillary axis without calibration are required for this project. A bonus would be if they include a calibration –free mode that estimates the visual axis and thereby the gaze point, as does the eye-tracker used in the second generation device.

These are just a few of the metrics that need to be kept in mind while choosing hardware for the next generation device. Finally, the device cost is a very important consideration. Both prototypes developed so far have had a high cost, which would be unaffordable to the average public. With the DK2, the high cost was primarily because of the high cost of retro fitting eye trackers. Up until now, eye trackers have always had to be custom-fitted into the HMD as an aftermarket upgrade. As eye tracking gets more popular, eye trackers will become part of the HMD device bought off-the-shelf. HMDs with built-in eye trackers will be orders of magnitude cheaper than those which have to be custom-fitted.

113 FOVE is a company supported by Microsoft Ventures that has been working on creating the world’s first VR Eye-Tracking HMD. The device being developed has the following relevant specifications:

• Weight: 400g

• Display FOV and Resolution: 100+ degrees, 2560x1440 resolution • Built-In Stereo Eye Tracking: 120fps per eye, <1-degree accuracy

The waitlist for the developer’s kit has been joined, which will allow us to buy the device as soon as it hits the market. Since the eye trackers are built into the device, and the device is mass manufactured, the developer’s version is quoted to cost around $600. This is a huge improvement when compared to the ~$15000 price tag of the second generation prototype developed. FOVE’s device is similar to the Rift DK2 in that it would need to be powered by the computer. It is recommended that this device be purchased when the opportunity arises.

SMI recently released their eye-tracking upgrade for the GearVR used in the IRIS Vision device described previously. The GearVR uses a phone as the screen, processor, and camera as described previously. Specifications of the device + upgrade are as follows:

• Weight: ~350g

• FOV: 96 degrees. Resolution: ~2560x1440, depends on phone used

• Eye-Tracking upgrade: 60Hz binocular eye tracking, 0.5-degree accuracy. Offers calibration-free mode, 1, and 3-point calibration similar to the current device.

114 The base model of the device costs $12000, but with the analysis software included costs $32000. The advantage of this device is that potentially no laptop is required, as the phone could be used for the remapping. This would be a step towards are completely portable device. However, it still needs to be investigated if the phone has the processing power to carry out the remapping in real-time. Finally, since programming for the phone would need setting up a new software framework, this would be a time consuming process. With this in mind, it is recommended that this device isn’t explored until testing on AMD subjects is concluded with positive results.

Other companies and devices to look out for are the Oculus Rift CV1 and the HTC Vive. These devices do not contain eye trackers built in. Thus, simultaneous monitoring of the aforementioned eye-tracking companies for any eye-tracking upgrades for these devices would be required. To summarize, it is recommended that devices that improve upon eye tracking and HMD specifications be considered for the next generation prototype, with a goal of improving user experience. Improving portability should be prioritized below improving user experience, and portable devices on different platforms (such as SMI + GearVR) must be explored only once a fully-functional computer-based device is developed and tested for efficacy.