Google Project Tango 资料简介(一)

一 软件:

C++版代码:https://github.com/googlesamples/tango-examples-c

Java版代码:https://github.com/googlesamples/tango-examples-java

Unity版代码:https://github.com/googlesamples/tango-examples-unity

SDK下载:https://developers.google.com/tango/downloads

开发者网站:https://developers.google.com/tango/

API:https://developers.google.com/tango/apis/overview

二 硬件:

第一代:

五寸手机平台Peanut,深度信息检测是基于Primesense structured light方案,通过第一代Tango设备的开发,Google基本完成了Tango Core和SDK的框架搭建,也初步验证了Tango的整体空间感知能力,但由于采用了结构光的深度检测信息技术,导致系统的尺寸、重量和散热都无法做到一个真正的手机中去。

第二代:

7寸的平板电脑YellowStone,基于NVDIA Tegra K1处理器和Mantis Vision的structured light深度检测方案,由台湾仁宝代工,虽然这款平板号称上市,但却是限量发售,并且只面向开发者,普通消费者是无法购买的。无法作为第一款商用的原因有二: 首先此款平板电脑由于采用了Tegra K1处理器和Mantis Vision的视频处理芯片,功耗其实比第一代还要严重;其次由于Primesense的机构光技术被Apple收购之后,将无法继续用结构光的深度检测技术方案,很明显不适合大批量的生产。

具体参数如下:(关键参数标红)

Screen

7.02" 1920x1200 HD IPS display (323 ppi)

Scratch-resistant Corning® glass

Dimensions

119.77 x 196.33 x 15.36mm

Weight

0.82 lbs (370 g)

Cameras

4 MP 2µm RGB-IR pixel sensor

1 MP front facing, fixed focus

OS

Android™ 4.4 KitKat®

Wireless

Dual-band Wi-Fi (2.4GHz/5GHz) WiFi 802.11 a/b/g/n

NFC (reader and peer to peer modes)

Cellular connectivity: LTE, HSPA+, 3G, 2G, GSM, EDGE

LTE: Bands 2,4,5,7,17 (1900, 1700, 850, 2600, 700)

HSPA+: Bands 1,2,4,5 (2100, 1900, 1700, 850)

Note that CDMA is not supported.

Note also that the TDK has not been certified by any mobile carriers, so functionality and performance with any particular carrier is not guaranteed.

Audio Output

Dual stereo speakers

3.5mm audio connector (OMTP standard)

Memory

128 GB internal storage (actual formatted capacity will be less)

4 GB RAM

Ports

Micro HDMI

USB 3.0 host via dock connector

Micro SD card

Nano SIM slot

Battery

4960 mAH cell (2x 2480 cells)

Processor

NVIDIA Tegra K1 w/ 192 CUDA cores

Sensors

Motion tracking camera

3D depth sensing

Accelerometer

Ambient light

Barometer

Compass

GPS

Gyroscope

Certifications

North America: FCC/IC, CEC BCS, UL, PTCRB

Worldwide: CE, CB, GCF, GCF FT

Battery: UL, UN, CB, CE

Adaptor: UL/ULc, CB

更多参数:https://developers.google.com/tango/hardware/tablet

第三代:($499)

联想一起发布的Phab2 Pro智能手机,Tango部分的整体构架弃用结构光,第一次采用TOF深度检测技术,在整体的功耗、整机尺寸和Tango Core上都做了深度地优化

具体参数如下:(关键参数标红)

Screen 6.4" QHD (2560 x 1440) 2K wide-screen IPS assertive display 2.5D curved glass

Design 3.49" x 7.08" / 88.57 mm x 179.83 mm aluminum unibody Weight: .57 lbs / 259 g Fingerprint scanner (rear side)

Cameras: Rear 16 MP RGB fast focus Depth sensor Motion tracking sensor / fisheye

Camera: Front 8 MP fixed focus, F2.2 aperture, 1.4 um big pixel

Processor Qualcomm® Snapdragon™ 652 processor built for Tango

SIM: Dual slots Nano SIM & microSD (Up to 128 GB)

Battery* 4050 mAh Li-ion + fast charge Standby up to 13 days Talk-time up to 18 hours

Memory 64 GB ROM 4 GB RAM Expandable memory up to 128 GB, microSD card

Audio Triple array microphones with active noise-cancellation Dolby Atmos® / Dolby Audio™ Capture 5.1 3.5 mm audio jack

Operating System Android 6.0, Marshmallow

Connectivity Speed: 2.4 GHz & 5 GHz WLAN: 802.11 a/b/g/n/ac Bluetooth® 4.0

Colors Champagne Gold (Coming Soon) Gunmetal Grey

Preinstalled Software* McAfee® Security Netflix SwiftKey

更多参数详见:http://shop.lenovo.com/us/en/smartphones/phab-series/phab2-pro/

三 关键技术

1.运动追踪(Motion Tracking)

在传统9DOF(惯性测量单元、加速度计和陀螺仪)的运动检测基础上,额外增加了超广角鱼眼Camera,通过超广角鱼眼camera的视觉特征识别补偿,从而消除单纯IMU运动轨迹的累计误差。相当于SLAM的特征检测环节。具体过程为:拍摄——识别特征点——匹配特征点——筛去错误匹配——坐标换算。

更多参见:https://developers.google.com/tango/overview/motion-tracking

2.深度检测(Depth Perception)

第一代Peanut采用Primesense结构光方案,第三代联想Phab2 Pro所采用的TOF技术。Project Tango的深度检测是实时的立体空间检测,通过Tango Core优化的算法,配合Motion Tracking超广角鱼眼Camera抽取的特征点,快速地实时把实际环境动态建模,从而避免了海量运算,这也为移动设备能够集成Tango提供了可能性

更多参见:https://developers.google.com/tango/overview/depth-perception

3.区域学习(Area Learning)

Tango设备检查到的空间深度信息和超广角Camera检查到的特征点匹配成一组场景信息存在Tango设备中,随时跟检查到的新场景(深度和特征点)做匹配运算,一旦发现检查到的新的场景和某个预设的场景中深度和特征点信息匹配,则Tango设备会把自己重新马上定位到预设的场景中去,并且同步修正追踪的数据,为下一次场景匹配做准备,终而复始地做如上的操作。相当于SLAM中的闭环检测环节。

更多参见:https://developers.google.com/tango/overview/area-learning

你可能感兴趣的:(Google Project Tango 资料简介(一))