new chinese gumstix competitor 74 USD AllWinner A10 Android

[limor]

http://www.cnx-software.com/2012/05/17/ ... dmi-stick/
good for robotics.
low cost small form CPU (maybe?) with floating point.

AllWinner A10 @ 1.5GHz + Mali 400 GPU
Memory 512MB RAM
Storage 4GB Flash
microSD slot (Up to 32GB)
Connectivity WiFi 802.11b/g
USB micro USB 2.0/OTG port
USB 2.0 Host port

The device can be purchased on Aliexpress for 74 USD (including shipping) where they say it’s model MK802 or on Alibaba for 368 RMB (~58 USD). Considering the Mele A1000 price is now around 90 USD (inc. shipping)

[limor]

and another one:

NEWEST Android4.0 IPTV C77 Google HD Player MINI PC Google TV Smart box Telechips CORTEX-A5 512M 4G (Smaller than MK802)

http://www.aliexpress.com/wholesale?Sea ... 77+android

[limor]

if anyone comes across info regarding floating point performance of these modules, please post here.

For hobby robotics we need some new metrics for these embedded boards.
What do we actually need all this computation power for?

1) Linux, WiFi and USB
-- connect an onboard camera
-- connect an onboard microprocessor that controls the motors and reads sensors (arduino or not)
-- any non linux (or windows/android) environment is too much hassle for hobby purpose

2) Enough CPU power (some floating point needed)
-- path planning (process sensors from microcontroller and control motors)
-- comprehensive gait generation, inverse kinematics and kinetics (for legged robots)

3) Vision processing (lots of floating point needed)
-- either relay the video over wifi to a remote PC and use Roborealm
-- or have enough floating point power onboard
-- There's either Intel or ARM. If we can afford Intel Atom then we can do pretty much anything on-board and throw in a kinect/xation for in-door depth vision and mapping

Assuming we go the route of onboard vision processing, what vision wizardry do we need for our robots to do clever things and at what resolution and frame/sec do we need those to perform?
I'd like to see a price/performance chart for robotics on-board CPU boards useful vision applications benchmark (google "robot vision benchmark")

- video face detects [VGA resolution] per second (robot looks at person)
- points of interest detection [VGA resolution] per second
-- vision based localization
-- object tracking

[nicolas gomez offline]

Hi

in this links a comparative pdf


http://omgfire.narod2.ru/SBC_comparison39.pdf

http://www.cnx-software.com/2012/06/26/list-of-39-low-cost-linux-friendly-boards-and-products/

[limor]

thanks for this comparison list
Great work!

floating point / opencv benchmark is what we need next

[limor]

I bought this for $40 including shipping after finding a discount coupon online (google)
http://www.pandawill.com/cx01-mini-andr ... 63620.html

it's been a couple of weeks but today they said they will send it out "soon"

can't wait to test some image processing and maybe incorporate it into our linuxified.net framework

[nicolas gomez offline]

I bought this for $40 including shipping after finding a discount coupon online (google)
http://www.pandawill.com/cx01-mini-andr ... 63620.html

it's been a couple of weeks but today they said they will send it out "soon"

can't wait to test some image processing and maybe incorporate it into our linuxified.net framework

Hii limor

here Linux Kernel 3.0.8 and Android ICS Source Code for your new mini-android

http://www.cnx-software.com/2012/07/17/ ... urce-code/

[limor]

thats great news

any pointers to how to create the toolchain for a Linux distro ?
I prefer not to use Ubuntu because this little device will never be plugged into HDMI beyond the first installation and testing. It will only be running robot brain activities so I prefer that the O/S will be running the bare essentials. WRT may even be an option.

[nicolas gomez offline]

thats great news

any pointers to how to create the toolchain for a Linux distro ?
I prefer not to use Ubuntu because this little device will never be plugged into HDMI beyond the first installation and testing. It will only be running robot brain activities so I prefer that the O/S will be running the bare essentials. WRT may even be an option.

I just thinking try use qnx or linaro
https://www.miniand.com/forums/forums/2/topics/82

http://software.intel.com/en-us/article ... -or-linux/

but need search more about qnx use

[limor]

long overdue update.
Tried to use this little device.
(which sells on Amazon now for 29GBP)

But WiFi trickles at a rate of 1 byte per second.
totally useless without wifi.

It also needs 500ma (ie: laptop output or usb powered hub) otherwise the screen keeps going dark every few seconds. It also heats up so much it can barely be touched.

I guess I will have to install the alternative firmware. i was hoping to actually use this module to replace the old PC sitting behind my TV. seems like it will have to be used as robot brain instead.

[limor]

looks like many have experienced the same problem

http://www.pandawillforum.com/showthrea ... ance/page3

[creed10 offline]

I hope you will keep posting nice stuff.

[kasierchiefs offline]

Hi! It is kind of a late comment, but has it been an update lately? I'm interested but I´m not sure if it has already been upgraded.

[nicolas gomez offline]

HIi all!!
here tha last update.

A80 OptimusBoard - AllWinner Octa-Core , SoC – AllWinner Ultra Core A80

CPU
4x Cortex 15, 4x Cortex A7 At 2.0Ghz

GPU
Imagination Technologies PowerVR GPU 64-core G623

System Memory – 2GB DDR3
Storage – 8GB NAND Flash
Video Output – HDMI 1.4a
Connectivity – Dual band Wi-Fi 802.11 b/g/n, Bluetooth 4.0, and Ethernet (GMAC)
1x USB 3.0 OTG HIC
2x USB 2.0 host ports
Audio codec – AC100 Codec

Camera
Camera I/F – 12MP MIPI CSI
Integrated parallel and MIPI I/F sensor
Supports 5M/8M/12M/16M CMOS sensor
Supports 8/10/12-bit YUV/Bayer sensor

Debugging – UART and JTAG
Expansion – 32-pin GPIO header
Misc – IR receiver, reset and power LEDs.
Power management – AXP806, AXP809
Power Supply – DC IN (5V), and battery

Size – 135x70mm
be available in May-June for less than $100.
Here a video
https://www.youtube.com/watch?v=et0tUD_mL0E

http://www.allwinnertech.com/en/developer/board/




Allwinner A80, with USB3.0 support (faster transfer and charging), Allwinner A80 Optimus Board, 4K H265 Encode and Decode in real-time, talking performance/power advantage over competition, 64-core G6230 GPU, Allwinner for car safety.
Support Opencl 1.X,Opengl 3.x, Direct 3D, Renderscript.

Here a more complete review about Dev-boards - GPUs - mini pc- SoM modules
https://forum.poppy-project.org/t/vision-system-for-poppy/92/5?u=niko2009