afterimageproject

[afterimageproject]

Progress Report Week 10 (Winter)

The Pandaboard stopped powering after we had plugged in the Trainer-XM even though it is a listed as a Pandaboard accessory from the Pandaboard site: http://pandaboard.org/content/resources/Accessories .  

We managed to ordered a Beagleboard-XM overnight.  

GPIO interrupts have been implemented for the Atmega to communicate with the Beagleboard.

On Android, the sqlite database has been added and messaging format was finalized for the demo.

[afterimageproject]

Progress Report Week 9 (Winter)

We switched our board to the Pandaboard and successfully patched the UVC driver to lower the bandwidth requested by each usb web cam.

Additionally, the google maps api was implemented to the android UI to display the Afterimage location. 

The GPS module was successfully integrated into the system: we can now send the user an SMS message containing the car's location.

MMS was looked into in more detail. We are able to encapsulate an image into MMS format using python-mms.

Our posterboard for the design review was created.

[afterimageproject]

Progress Report Week 8 (Winter)

After testing our GPS module, we believe we have provided a non level translated transmit to our Em406-a module.  We purchased a usb-to-serial programmer to test the GPS module, and we do not receive NMEA sentence outputs.  We have ordered a GPS module from microcenter to pick up locally.

Additionally, the Beagleboard stopped working when we plugged in our GPS module's vcc to our i2cvcc on the trainer-xm with the jumper JP1 set to 5v.  We are uncertain if the trainer-xm was damaged in the process.  We will use our spare Pandaboard in the mean time.

[afterimageproject]

Progress Report Week 7 (Winter)

More testing was done at night with the webcams we had purchased this week. The camera with built-in IR LEDs had sufficient lighting at night, but very poor resolution. The camera without the IR LEDs had much better resolution, but did not work well in the dark.

The bug for with the accelerometer was finally fixed.  We discovered the data type int was 16bits vs the expected 32-bits and this caused the accelerometer to work sporadically in our tests. 

Created working code to trigger an SMS text message when the accelerometer detects a crash (via GPIO polling).

[afterimageproject]

Progress Report Week 6 (Winter)

This week, the user android application was written.  Currently it has only basic functionality - it sends a message to the Afterimage system.  Settings, UI, and receiving functionality has not been completed.

We programmed the atmega328p for reading voltages (for the accelerometer).  We are currently debugging issues.

The SMS sending program was prototyped for Beagleboard.  We are still looking into sending images via MMS using our GSM module. 

Other webcams were purchased that were UVC compliant for testing.

Gstreamer application development is still being researched.

[afterimageproject]

Progress Report Week 5 (Winter)

Sending and receiving messages, and other functions we are planning to use were tested via usb on the GSM module using AT commands.  We are currently looking into sending images as AT commands do not seem to support this or MMS in general.

Tested methods of obtaining video from initial usb camera (VLC, Gstreamer, FFmpeg, etc). ��Purchased 2 different usb cameras are being purchased for additional testing.  Additional cameras will be purchased after comparing the two cameras.

Code for the GPS module has been written, but we are purchasing an additional breakout for the module.  

[afterimageproject]

Progress Report Week 4 (Winter)

Due to time constraints, we are using a usb interface for the webcam for now.  Using the camera connector just to move raw video data required more knowledge of writing drivers and the v4l2 kernel level drivers than originally anticipated.

We completed setup and testing of the trainer-xm's Atmega328p.  The accelerometer module was then tested and coding has been completed.  The PIR sensor was tested.  Using ffmpeg and v4l2, the creative live webcam was not working.  This may be a webcam compatiblility issue. With VLC, the webcam was able to stream and record video.

[afterimageproject]

Progress Report Week 3

The ov7670 v4l2 driver uses i2c interface to communicate with the camera module;  however, the pandaboard did not have a 8-bit data connection with sufficient speeds to handle the camera data coming in.  The camera interface on the pandaboard uses CSI-2 specification which is not released to the public.

Beagleboard on the otherhand, has a camera interface which will accepts data 0:11 at a 24mhz clock.  We are currently ordering a Beagleboard due to the data transfer situation.

[afterimageproject]

Progress Report Week 2 (Winter)

When the power adapter arrived week 1, a chip on the Pandaboard fried after plugging it in 5v 3.6a.  During the previous week, a 12v power adapter with the same 2.1mm barrel connector was plugged in for a few seconds triggering the overvoltage protection led.  Unfortunately, the overvoltage protection does not work as other users have experienced similar issues after using power supplies greater than 5v.

We tested the new Pandaboard with the validation environment to ensure the hardware was working with the image provided by pandaboard: http://pandaboard.org/node/13531/ .

Next, we loaded our Ubuntu 11.10 ubuntu image onto the sd card.  Our next step will be adding i2ctools to our environment, which provides register level access helpers, to test reading the device id of our camera module.  We ordered a header for the Pandaboard expansion port, and plan to solder wires to the camera modules.

[afterimageproject]

Progress Report - Week 10

After discovering interfacing with the cmos image sensors was not going to be viable due to the size of the pins, we looked into purchasing image sensor modules.  We decided on the omnivision OV7670 image modules.  

We also discovered the issue would also carry over to the GSM module we planned to purchase, but located a breakout board available for the module.

[afterimageproject]

Progress Report - Week 9

We ordered several of the main components including the Pandaboard, Trainer-XM (extension board to add GPIO on the Pandaboard), image sensors, and other parts. Additionally, we worked on the third milestone report which included organizing our schedules, setting deadlines, etc.

We also looked into solutions for batteries and power. The implemented automobile must provide our module with power. However, this can be rather tricky since being on an automobile implies that there are no convenient outlets available. 

[afterimageproject]

Progress Report - Week 8

Motion Detector Research: 

PIR sensors:

http://www.ladyada.net/learn/sensors/pir.html 

http://www.e-ele.net/DataSheet/BISS0001.pdf 

  takes the analog output of the PIR, and processes it into a digital pulse. (Needed if we want to built the sensor ourselves).

Typical range is 20ft, 100 degrees horizontal, 60 degrees vertical.

 Things to consider:

Cars that have already been running for a significant amount of time will be detectable (engine/brake heat). However, cars that have been parked for a while might cause a problem.

If the sensor is turned off, it can take up to a minute to power back on and must calibrate without IR disruption.

Will have difficulty in detecting motion coming directly towards the sensor.

  http://www.mouser.com/ProductDetail/Parallax/555-28027/?qs=sGAEpiMZZMvhQj7WZhFIAGvhT1do5x5QYCQRbP48WHs%3d

$10

Two modes of operation: long range and short range. Long range detects motion up to 30 ft, short up to 15ft.

Assuming we can get the system to wake from hibernation quickly enough, a shorter range mode might be better since less cars/people will turn on the system (lower overall power consumption).

1.41 x 1.0 x 0.8 in

http://www.adafruit.com/products/189

  $10

~22 ft operating range with a 120 degree cone

http://www.mouser.com/ProductDetail/Panasonic-Electric-Works/EKMC1603111/?qs=sGAEpiMZZMvhQj7WZhFIAJaM7IUEh%252bHgM6uLDBL6vt0%3d

  $12

Range: 40ft, 102 deg. horizontal, 92 deg. vertical

Accelerometer Research: interior detection: Use motion sensor? X10 EAGLE EYE MOTION SENSOR MS14A http://www.ebay.com/itm/X10-EAGLE-EYE-MOTION-SENSOR-MS14A-OUTDOOR-INDOOR-/400151903889?pt=LH_DefaultDomain_0&hash=item5d2ae97e91#ht_1425wt_936 - cheapest on ebay $13. - integrates only to x10 system motion detecter component Pyroelectric Infrared IR Motion Sensor Detector Module http://www.ebay.com/itm/Pyroelectric-Infrared-IR-Motion-Sensor-Detector-Module-/390363005500?pt=LH_DefaultDomain_0&hash=item5ae372c23c#ht_3036wt_1170 - Infrared sensor with control circuit board. - $3 - ship from Hong Kong. May take up to a month to receive. - 7m range detection - working voltage range: 5-20v Pyroelectric Infrared PIR Motion Sensor Detector Module http://www.ebay.com/itm/Pyroelectric-Infrared-PIR-Motion-Sensor-Detector-Module-/200654955273?pt=AU_Fragrances&hash=item2eb7f7a309#ht_4023wt_1177 -same as above, can work at 4.5v ultra low power motion detector (paper) http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=5373819 etc: http://telematicsnews.info/2010/10/06/vw-toureg-gets-valeos-360vue-system_01061/ Valeo’s 360Vue system 360 view from above by processing image from 4 cameras on car

[afterimageproject]

Progress Report - Week 7

Cameras Researched:

There are 3 major requirements for a camera in this project: the ability to take clear photos of the subject in varying brightnesses, output that correlates with our needs, and a reasonably low power consumption during standby.

Will need camera of minimum 480 horizontal resolution, less than one megapixel.  This requirement should be simple to fulfill.  Should have max fps of 30

For sensitivity, the camera should have a minimum scene illumination of at most 1 lux for external cameras and .1 lux for internal cameras.

Camera should have some type of raw output so that we can write a driver.

Ideally, camera should have drivers available so that we can use those in case of emergency.

Our system will spend a majority of time in standby mode.  This means the standby power consumption of our camera, or boot speed, should be minimized

Based on the requirements above, a CMOS camera, rather than a CCD, would be ideal for our project because it is generally cheaper, and has greater sensitivity.  Although it generally has lower resolution, our resolution requirements are very low.

Pixel Size is a factor that we would consider when choosing between the final candidates of our project.  Although greater pixel size causes greater dynamic range and less noise, it also means a lesser resolution per the same unit area.  Considering that we are trying to install concealable cameras, this could become an important factor.

The final deciding factor is simplicity.  Ideally, we would like to purchase the image sensor and lens separately and install it ourselves, but this seems to be too complex for our needs.

[afterimageproject]

Progress Report Week 6

This week, we worked on camera and board comparisons.  We created charts and decided on which components we would use and why.

The board comparison: https://docs.google.com/document/d/1ReKcwSdNVSR7lstVryHJYySkBGx3zCXQ57iQ0zBqs-Q/edit Board comparison Chart: https://docs.google.com/spreadsheet/ccc?key=0AqZf_qL56WC8dFdfeXY0TkhVVmtTeU9MQlA3VlpxM1E Camera comparison Chart: https://docs.google.com/spreadsheet/ccc?key=0AqZf_qL56WC8dDJ1REQzaUZ2ejgzNC1Na3lVUUdSZkE

[afterimageproject]

Progress Report - Week 5

Research:

image/video compression, raw image data size, plausibility for real-time compression with parts and constraints of the project.  

potential parts to use for our project (Beagleboard-XM, OV7225 image sensor, ).  Further research and analysis needs to be done on the necessary system components.

the average MMS size limit (~300kB) to send data over GSM was found to be too small to send video for this project. One possible alternative would be to use TCP/IP protocol over GPRS to send the video in packets. A likely better solution would be to send pictures over MMS immediately, and store the video locally in the car's "black box".

[afterimageproject]

Progress Report - Week 4

Progress Report - Week 4

Design/component research: OnStar customer complaints

http://www.consumeraffairs.com/automotive/onstar.html

common complaints: 

customer service would not unlock car at customers' requests: process too secure and unnecessary

does not promise advertised: customer service would not track stolen vehicles without police report, advertisement states that tracking would start immediately

OnStar module too integral: if it breaks down, car cannot be started 

Learning from OnStar's complaints, we know to design a system that is not too central to the car. If the system stops working, the car can still operate. Car tracking needs to start immediately when stolen. A possible unlocking mechanic needs to be secure, but not unnecessarily complicated for the user. 

similar projects/products

1) ZBlackbox-1000

 0.1 LUX ultra light sensitive 1/5 CMOS sensor

recorded at 30fps at 656x488 resolution

Adjustable recording options - records up to 60 seconds before accident and up to 600 seconds after the accident

2) http://www.michigan-sportsman.com/reviewpost/showproduct.php?produc

Takes pics at adjustable resolutions

5 MP takes vid at adjustable resolution 20fps at 320x240, 640x480 (16fps)

uses 5 MP CMOS cam, cost can run up to 600

uses GSM

trigger speed is 1.2 s

3) http://www.google.com/products/catalog?hl=en&q=mini+on+board+cameras&cid=10337467865263952093&ei=9sKfTqqAGZDQNIXwrOIK&ved=0CAwQrRI

2 1.2 MP Cameras

4 infared LED

G-shock sensor

4) http://www.brickhousesecurity.com/voyager-car-camera-gps-logger.html

IR night vision

storage on device

5) http://www.alibaba.com/product-gs/320168852/gsm_car_security_camera_SP_5876.html

GSM security camera

night vision

Connection Method

Wi-fi hotspot - using a stand-alone wi-fi device

GSM modem  - need to buy SIM card

ADH8066 GSM Module

Evaluation board (optional)

GSM antenna

4595 3G Mobile Broadband USB Modem

Microcontrollers

 1) Digital Media System-on-Chip (DMSoC)

http://www.ti.com/product/tms320dm355

2) OMAP35x Applications Processors

http://www.ti.com/dsp/docs/dspcontent.tsp?contentId=53403

3) Beagleboard 

http://beagleboard.org/hardware-xM

HD-Video capable, usb, ethernet, etc.

4) Leopardboard

http://designsomething.org/leopardboard/default.aspx

5) Arduino

http://www.arduino.cc/

Economical, but probably too slow for video processing

Potential cameras

Considerations: What if it is night time? IR? Night Vision?

http://www2.cop-usa.com/ebproductdetail.asp?id=958 

http://www.securitycamera2000.com/products/600TVL-SONY-SUPER-HAD-CCD-Color-Mini-HD-Board-Camera-Wire-Control-OSD.html 

http://www.jameco.com/webapp/wcs/stores/servlet/Product_10001_10001_2144340_-1 

http://www.jameco.com/webapp/wcs/stores/servlet/Product_10001_10001_171029_-1

[afterimageproject]

Group Members

Here are our group's members: 

David Guo

David Lee

Matt Livezey

Yiyang (Matthew) Yan