Operating System: Debian sid
Versions:
- postgres 8.3.1
- postgis 1.3.3
- pgRouting1.02
1. Install base system and ssh
#vi /etc/apt/sources.list to include
deb http://debian.cn99.com/debian etch main
deb-src http://debian.cn99.com/debian etch main
#apt-get update
#apt-get upgrade libc6
2. Install the required packages for postgres8.3 and postgis1.3.3
#apt-get install sudo nmap telnet
#apt-get install python2.5 python2.5-dev python-setuptools
#apt-get install g++
#apt-get install build-essential cmake ibboost-graph-dev
#apt-get install libreadline5 libreadline5-dev
#apt-get install zlib-bin zlib1g-dev
#apt-get install libkrb5-dev
#apt-get install libcurl3
#apt-get install libssl-dev
#apt-get install postgresql-8.3
#apt-get install postgresql-8.3-postgis
#apt-get install postgresql-server-dev-8.3
3. Installing pgRouting
# tar -zxvf pgRouting-1.02.tgz
# cmake .
# make
# make install
Been very busy for the past few days/weeks in the office. Just read this one from Google Reader, Nokia develops navigating system based on image recognition, landmarks. In summary, base from Nokia’s Palo Alto Research center, providing direction guiding varies from gender and culture. However, providing examples such as “Turn Right after the next gas station” seems to be a universal and a normal behavior of providing directions, the person can relate more to the advice than providing line routes above synthetic maps.
This is interesting facts as it supports our theory in Call Center Directed Guiding. Although, a 360 or panoramic view of any point along the line would be ideal, an easier and more realistic approach for the time being would be to include important landmark snapshots along the route. However, I believe our organisation’s project is one step ahead since we have a warm body who already gave a summary of the route during the conversation. This means the caller have an idea of the route complexity beforehand. Then, we could also send this landmark images to the subscriber’s mobile as his future reference (which means additional VAS).
Another interesting location technology, fruitful for direction guiding, would be iPointer. I believe, base on the coordinates and orientation, the server could algorithmically predict which POI you are looking at or infront of. They mentioned they have a thin client on the mobile device, so I guess the S60 client tries to get the orientation base on the bearing. Also check this one out, “users can now walk down the street anywhere in Japan and point at over 700,000 objects such as buildings, shops, restaurants, banks, historical sites and instantly retrieve information on what they are looking at or find what they are looking for just by pointing their phone.”. Niche eh? Just compells me more to start up our mobile pursuit…
But what rocked me the most was indoor tracking by Ubisense using ultrawideband technology which is close to a foot?!?. This I have to see, as it opens new business cases for direction guiding indoors such as big malls, etc. Imagine mapjack views inside malls with Ubisense technology. Sweet.
Other References:
http://www.openlandmark.com/
Im trying out ways to create a topology of edges and vertices for a road network. Currently, there are three ways I know of:
1. Using ArcGIS build coverage line.
– This includes the use of ArcGIS. Exporting the feature into tics, arcs and nodes, then afterwards assembling them all together. Its functional but have not fully tested the quality of the road topology. Also, assembling them back together through the spatial objects comparison will take some time. It would be better *If* ArcGIS could create the nodes wrt to the base table.
2. Using PostGres, postlbs functions.
– Use of SELECT ASSIGN_VERTEX_ID(‘table_name’, double_precision_distance. So far this bails on me on my first test on my win-xp laptop. We can test this on Linux if the response is the same.
3. Using Mapinfo Basic Scripts provided by J.
– Haven’t gone indepth with these for now, but looking forward to it.
4. Using Grass
– As documented in pgrouting.postlbs.org site’s Topology Creation