Roadloc is a road geodata loggerand is used to record both human and instrumented readings during a road surveyusing an car.
The human observations include:
- road condition using standard engineering terms
- severity level of any damage to the surface observed
- risk to motorists
- comfort level
instrumented readings include:
- Accelerometer to derive an index of comfort
- Video Camera
The results of each survey feed into aroad network model for the whole region or state and the data is used for bothstrategic and tactical planning of a road network.
An example is that it helps providemore accurate data for road maintenance.
Roadloc can be used by local, state ornational construction authorities or their contractors so that everyone is speaking a common language and reference for the raod network.
I developed Roadloc Mk1 31 yearsago and it is still used today but augmented by GPS measurements.
MK1 was designed using the world'sfirst notebook computer, an Epson Hx-20 and it has made a major contribution tothe roads program in NSW in Australia. Car odometer measurements were feed into the bar-code interface of the HX-20 and data records were indexed against theseodometer readings. Mk1 transferred data using RS232 to a large mainframecomputer to create a road GIS model for the whole of NSW in Australia.
Why is Mk2 better?
- Mk1 used both a voice recorder and a keyboard so the observer was very busy just with these alone with not much time to look at the road. The observer had to look at the road & enter data into a keyboard
- Mk2 only uses voice commands instead of a keyboard and data input from sensors
- Mk1 stored the voice messages in analog form onto separate micro cassettes. Data was stored on microcassettes, printed records and analog voice records
- Mk2 translates the spoken observations into text using key words and this is stored into a database. There is no extra handling of the storage media
- Mk2 uses either standard GPS or differential GPS for a more accurate location so lane position can be identified and recorded
- Mk2 also measures ride comfort
- Mk2 has a video camera on the dashboard
- embedding GPS readings into MP4 video
- use GPS as a index to video frames
- accurate Speech to text and translation to key words with 95% accuracy level and minimize the training needed for new observers
- start initially with GPS and process latter to dGPS
- instrumented measurement of comfort level using current standards
- Microcomputer - LinkIt as a front end to the notebook computer
- Notebook Computer in the car communicates to both the corporate computer and the microcomputer
- Sensors: GPS or dGPS, accelerometer and video camera
- Database: mySQL
- sensor processing: node-red modules
- Speech to text: IBM or google 'Now' APIs
- GPS processing
- Software IDE such as Charm and node-red on a Chrome browser
How do I know it will work?
Testeach stage of the project
- Does the text to speech provide accurate data?
- Is an accelerometer sufficient to measure ride comfort?
- Can frames on the video be accessed from an GPS index?
- Is it easy for the human observer to use? What % time is spending on observing compared to using the computer?
- Can the survey records be readily transferred to a corporate database?
Has the final result been achieved andis Mk2 better than Mk1. Does it produce more accurate and reliable data?