DIY Optic Flow Sensor for autolanding (using a cheap Optical Mouse)

[Bg~]

Hi all..I recently attended a AIAA conference on UAVs that included a paper by a BYU group who used an optical mouse sensor to measure optic flow. This in turn was used to estimate height above ground for an autolanding routine by comparing to GPS groundspeed and barometric pressure.

The mouse sensor used was the Agilent 2610 sensor (information here (http://www.home.agilent.com/USeng/nav/-536893734.536883756/pd.html)) which can be had from a cheap Logitech optical mouse ($14 from Walmart, or $5 from newegg.com)

http://bg1.us/pics/optic_sensor/package_tn.jpg (http://bg1.us/pics/optic_sensor/package.jpg)
(Click for larger)

According to the data sheet for the A2610, the sensor itself has serial output, which allows the reading of horizontal and vertical speed as measured by the sensor, image feature count, frame rate, etc. All of this is stored in registers on the sensor itself. The mouse uses a separate microcontroller to read this serial data, and convert it to USB.

I'd like to get some more information on from someone more knowlegeable in electronics than I on a basic idea of what I could do to get access to the serial output from the chip itself. Maybe another microcontroller with serial I/O could do this? I want to have at least horizontal and vertical speed, but am a newbie to electronic stuff like this.

http://bg1.us/pics/optic_sensor/cboard_tn.jpg (http://bg1.us/pics/optic_sensor/cboard.jpg)
(Click for larger)

The sensor will require a lens to function when focusing on a distant image. I haven't looked for any suitable lenses yet.

[Mr.RC-CAM]

The data sheet shows that it is a SPI interface. Some microcontrollers can provide that 2-wire connection in native hardware, where others will require you to emulate it in software in a process called bit banging. I doubt you'll have any trouble finding sample SPI code for your chosen microcontroller if you look around.

But, how in the world does this optical IC estimate height from an R/C model? Does the target landing area have big visual markers painted on it? Is there an online white paper that you can point us to?

RC-CAM

[Bg~]

According to the presentation I saw, the optic flow sensor is first calibrated by driving past a wall at various distances and speeds. For a constant speed, the optic flow reported by the sensor will lessen as the distance from the wall increases.

The system used by BYU included GPS, a pressure altimeter, and the optic flow sensor. By comparing the GPS ground speed to the amount of optic flow, you can determine your height above ground. The BYU group used this and the altimeter to produce an accuracte height above ground measurement. No visual markers were needed, although it was mentioned that a relatively featureless asphalt road caused problems, and above a certain height (40 meters I think) the system no longer worked due to the small resolution of the sensor (18x18 pixels).

The paper is not online at their site yet -- http://www.ee.byu.edu/magicc/

[william541]

RolfPW did some experiments with a similar device here, not sure if he did anything with it lately though.

http://www.rcgroups.com/forums/showthread.php?t=267651&page=3&pp=15

I can see where the "rate" sensor could give you ground speed if altitude was known. I guess it only stands to reason if true ground speed was known and the ground rate for a given veiwing angle was known, this should give you altitude. Kind of clever!

-Bill-