Project Tango and Visual-Inertial Localization

Project Tango and Visual-Inertial Localization

This project enti­tled “Large-Scale, Real-Time, Visu­al-Iner­tial Local­iza­tion” is inter­est­ing, using Google’s exper­i­men­tal ‘Tan­go’ hard­ware to improve real-time track­ing of loca­tion and position.

The hard­ware is a tablet com­put­er with a motion track­ing cam­era, a 4 megapix­el 2µp pix­el cam­era, inte­grat­ed depth sens­ing and a high-per­for­mance proces­sor. This equip­ment aids in tasks like scan­ning rooms. A lim­it­ed num­ber of kits were pro­duced and giv­en or sold to pro­fes­sion­al devel­op­ers with the intent of mak­ing tech­no­log­i­cal developments.

One day we may see more accu­rate and inter­est­ing aug­ment­ed real­i­ty. I’ve often thought over­lay­ing infor­ma­tion onto our cur­rent real­i­ty would be inter­est­ing. Walk­ing down a street and see­ing for-sale signs could be inter­est­ing. It may just being over­loaded in adver­tis­ing, mak­ing a vir­tu­al eye­sore though.

Source:

Get Out of My Lab: Large-scale, Real-Time Visu­al-Iner­tial Localization
Simon Lynen, Torsten Sat­tler, Michael Bosse, Joel Hesch, Marc Polle­feys and Roland Siegwart.
Autonomous Sys­tems Lab, ETH Zurich
Com­put­er Vision and Geom­e­try Group, Depart­ment of Com­put­er Sci­ence, ETH Zurich
http://www.roboticsproceedings.org/rss11/p37.pdf

Smart scanning technology detects early signs of potholes

Smart scanning technology detects early signs of potholes

Smart scanning technology detects early signs of potholesInter­est­ing work by a team led by Not­ting­ham Trent Uni­ver­si­ty on mak­ing a mobile scan­ner that can detect ear­ly signs of pot­holes. It uses con­ven­tion­al cam­eras, 3D scan­ners and com­put­er vision to detect ‘rav­el­ling’. Mov­ing this around at traf­fic speed mount­ed to a vehi­cle, and com­bin­ing it with GPS and a suit­able log­ging sys­tem, it could be used to improve road conditions.

Researchers are devel­op­ing smart scan­ning tech­nol­o­gy using exist­ing cam­eras to detect the ear­ly signs of pot­holes and deter­mine their severity.

The tech­nol­o­gy, devel­oped by a team led by Not­ting­ham Trent Uni­ver­si­ty research fel­low Dr Sen­than Math­a­van, scans roads for rav­el­ling — the loss of aggre­gates from the asphalt which leads to pot­holes and cracks.

Ravelling - an early indication of potholes
Rav­el­ling — an ear­ly indi­ca­tion of pot­holes — Source

Com­bined with 2D and 3D scan­ners on a pave­ment mon­i­tor­ing vehi­cle, a com­put­er vision algo­rithm can exam­ine the road with accu­ra­cy at traf­fic speed dur­ing day or night.

The sys­tem works by detect­ing dif­fer­ent tex­tures of the road to iden­ti­fy rav­el­ling and dis­tin­guish­es it from shad­ows and blem­ish­es such as tire marks, oil spills and recent pot­hole repairs.

“It’s imper­a­tive for author­i­ties across the world to be able to mon­i­tor road con­di­tions effi­cient­ly and safe­ly,” said Dr Math­a­van, a research fel­low of the School of Archi­tec­ture, Design and the Built Environment.

“For the first time, aca­d­e­m­ic research has addressed the issue of detect­ing rav­el­ling in an auto­mat­ed way, which has led to the devel­op­ment of this nov­el soft­ware which can be used across the industry.”

The research was pub­lished today in Trans­porta­tion Research Record, a lead­ing aca­d­e­m­ic jour­nal for trans­porta­tion infra­struc­ture research. It also involves Dr Mujib Rah­man of Brunel Uni­ver­si­ty, Mar­tyn Stonecliffe-Jones of Dynat­est UK Ltd, and Dr Khur­ram Kamal of the Nation­al Uni­ver­si­ty of Sci­ences and Tech­nol­o­gy in Pakistan.

Dur­ing the research, the team found that the tech­nol­o­gy detect­ed road sur­faces cor­rect­ly in all 900 images test­ed. It took approx­i­mate­ly 0.65 sec­onds to 3D process the rav­el­ling mea­sure­ments, but it is believed that this could be reduced further.

potholeDr Rah­man added: “Pot­holes, in their worst poten­tial form, can cre­ate dan­ger­ous dri­ving con­di­tions and cause cost­ly dam­age to vehicles.

“What this tech­nol­o­gy allows us to do is cap­ture bet­ter qual­i­ty infor­ma­tion on road con­di­tions, with­out dis­rupt­ing the flow of traf­fic or incur­ring unnec­es­sary costs.

“This could be a sig­nif­i­cant step for­ward in the way that pot­holes are man­aged, help­ing improve the time­li­ness and effi­cien­cy of repairs.”

Source: http://www.ntu.ac.uk/apps/news/169006 – 6/Smart_scanning_technology_detects_early_signs_of_potholes.aspx