Archive for March, 2012

Time differentials (repo update)

Saturday, March 31st, 2012

The Open­Shoe run­time frame­work and the Nav­i­ga­tion algo­rithm repos has recently been updated to use time depen­dent time dif­fer­en­tials. A clock reg­is­ter is read out on every inter­rupt and the result­ing time dif­fer­en­tials are used to adjust the “dt” in the mech­a­niza­tion equa­tions and the fil­ter. This we have found nec­es­sary if the IMU is free run­ning since the IMU oscil­la­tor has clock pace vari­a­tions (pri­mar­ily between dif­fer­ent units) of plus-minus 1%. Under some con­di­tions this gives sig­nif­i­cant per­for­mance degra­da­tion. The resent patch fixes this.

Project workshop at IISc

Friday, March 30th, 2012

Dur­ing the last two weeks a group from KTH (includ­ing me) has been vis­it­ing our col­legues at IISc and con­duct­ing a project work­shop related to the joint KTH-IISc Indo-Swedish project on indoor local­iza­tion. Dur­ing the work­shop, a current-status pre­sen­ta­tion of the Open­Shoe project was given. The pre­sen­ta­tion slides can be found at here.

Article pre-print

Friday, March 30th, 2012

A first ver­sion of the con­fer­ence arti­cle, which is to be pub­lished on the PLANS2012 con­fer­ence later next month, has been sent in for review. The arti­cle will not be final (no major changes expected) and avail­able here until in May. How­ever, if you are inter­ested in a pre-print of the arti­cle please con­tact us at openshoe@ee.kth.se.

2 more complete units assembled

Sunday, March 11th, 2012

After get­ting some more IMUs from Ana­logue Devices last week, tonight I assem­bled two more com­plete units. I was happy to see that every­thing worked fine out-of-the-box. How­ever, I man­aged to crack a cas­ing when I pressed down an IMU. The tol­er­ances of the IMUs are a cou­ple of tenth of mm and I guess this unit was slightly big­ger than the oth­ers. Be aware that this can happen!

Some small com­ments about new IMUs. First of all, per default they come with a 4-tap low-pass fil­ter turned on. This can be turned off with the set_lowpass_filter_imu.m file among the Mat­lab con­trol scripts. Also, out of the box, the gyros have rather large biases up to 1°/s. There­fore, the gyro­scope cal­i­bra­tion rou­tine should be run once to make the sys­tem per­form better.

Demo of the OpenShoe project at the IISc Open Day

Friday, March 9th, 2012

A demo of the Open­Shoe project” was held on the occa­sion of IISc Open Day on 3 Mar 2012 (also IISc Founder’s Day). It was held in the main foyer of the ECE Main build­ing. Sub­hadra, Kishore and Anand demon­strated the work­ing of the Open­shoe and wire­less video stream­ing in the con­text of the First-Responder sys­tem being devel­oped under the Indo-Swedish activ­ity. A hel­met, with a web­cam mounted on it, and the open­shoe was worn by Sub­hadra Singh and he walked around the foyer of the build­ing.
The video was streamed via the Bea­gle board using an ad-hoc Wi-Fi net­work to the com­mand con­sole at the entrance of the foyer which also dis­played the streamed posi­tion data in real-time.
Sev­eral hun­dred vis­i­tors (gen­eral pub­lic) enjoyed a thrilling expe­ri­ence of watch­ing the real-time stream­ing of tra­jec­tory and and video feed as the per­son walked around the foyer. A few pics attached.