Concrete Speech

Maintaining city streets and sidewalks is one of the most important functions of a municipal government. Regular inspection of city roads is an expensive, but necessary function of any city. In the Colorado Rockies, one city has come up with a speech recognition solution that allows them to perform the task better than ever, in a fraction of the time it took them in the past.

The Denver suburb of Aurora is home to 250,000 people and features 860 miles of streets over a 65 square mile area. Maintaining those roads requires regular inspection.

In most cities, the data required for inspections has to be collected by hand. Technicians survey the streets and sidewalks, then make notes on several different paper forms. When they return those forms to the office, data entry was done by clerical staff, assuming they can decipher the handwriting. This process takes days, and is subject to misunderstanding.

Recently, Aurora found a solution that involves speech recognition and voice-to-data capture techniques, the VoCarta System from Datria Systems. Technicians now use a mobile computer with speech recognition software. A Global Position-ing System (GPS) receiver tracks their location. Field data is collected from either a car or a specially equipped golf cart. Data is then downloaded into a host computer for analysis.

Aurora is not only performing the inspection task faster than ever before, the city is also increasing the type of data collected on the street network. So the system has improved both the quality of the inspections and allowed the city to conduct the inspection much quicker than by traditional means. Similar systems are being used in Austin, TX, Toronto, and New Brunswick, Canada.

Some examples of the type of inspection work being done by the Aurora Public Works Department include analysis of pedestrian handicap ramps and detailed concrete infrastrure.

Pedestrian Handicap Ramp Analysis

The Americans with Disabilities Act (ADA) has mandated that cities evaluate their curb and gutter installation for pedestrian handicap ramps at intersections. Each intersection in a city must be visually surveyed for pedestrian ramp installation. Aurora, which has a mixture of intersections with and without ramps, needed to survey over 10,000 intersections for ADA compliance.

Local consultants estimated that a conventional manual survey would take several months and cost over $30,000. Instead, Datria’s Vocarta system was used to accomplish this task with in-house technicians, in only eight weeks and at about half the cost. The voice to data collection system allowed information to be collected at close to normal driving speed.

Concrete Infrastructure

All cities have a significant investment in their concrete infrastructure. The condition of the curb, gutters and sidewalks in a city say a lot about the quality of life and image of the community. Cities with a failing infrastructure give the impression of a city in decline rather than prosperity. Every city also has cracked and damaged curbs and sidewalks which must be repaired each construction season. The extent of the concrete network, in most urban areas, is such that only the worst repairs can be completed each year.

Manual data collection methods have been used in the past in Aurora. The usual procedure was to collect data at each address location where concrete was in need of repair. The address was written down, along with a description of the repair type and quantity. The data was then entered into a computer database, and a repair contract was prepared for a local contractor.

VoCarta has dramatically changed the type, amount, accuracy and speed of data collected by technicians in the field for concrete repairs. Data collection speed has increased dramatically with fewer errors. Technicians can also collect information at a finer level of detail than they were previously able to collect using a manual system.

For the 1998 curb, gutter and sidewalk repair program, two technicians worked part-time over a two month period collecting data on over $1.3 million worth of concrete repairs. Contracts were prepared two months ahead of what had been the normal time frame. Completing this large amount of data collection in such a short time period would be impossible using conventional manual data collection techniques.

Network and Project Level Data

There are different classes of streets in the city, from major urban arterial streets carrying heavy traffic volumes, to residential neighborhood streets. Curbs, gutters and sidewalks are located next to the paved portion of the street in most locations. Condition data is necessary for both the paved driving portion of the street and the curbs and sidewalks next to the street.

Data on street and sidewalk condition is subdivided into two categories; network level data and project level data. Network level information defines either the roadway network attributes or the curb, gutter and sidewalk system attributes. Network level information is useful for planning, budgeting and scheduling. Project level data provides a more detailed inventory of the system attributes. Project level information is useful for determining the actual extent of cracking and provides field locations for repairs.

Network Level data includes an inventory of the following information: Pavement Network Information: Street Classification; Surface Type; Lane Count; Width; General Condition. Curb, Gutter, Sidewalk Network Information: Structure Type; Length; Location relative to Street Centerline.

The network level data is collected quickly by a field technician driving the roadway network in a specially equipped car. The set-up for data collection includes a portable computer equipped with headset, microphone and GPS antenna on top of the car. The technician drives down the road and verbally describes the network attributes into a microphone. A pre-selected grammar is used for either the pavement network or the curb, gutter and sidewalk network. The technician, talking into a microphone, uses some key words to describe the appropriate network attribute.

The voice recognition software records the words spoken by the technician during the data collection session. The GPS receiver attaches an accurate location to the asset described by the technician. The technician can collect data at approximately 20 miles per hour for the pavement network information and about 10 miles per hour for the curb, gutter and sidewalk information. After completing a collection session, the session is downloaded into an office computer.

The data is then transferred into an Arcview database. The data is then available for analysis using a custom work order system provided with Vocarta. Standard Arcview analysis tools are used to query the database for information.

Project Level data, which is much more detailed, is also collected. Project level information defines system attributes down to the crack level for a wide variety of parameters. The project level data is collected in a similar manner to the network level, but the speed of collection is slower because of the high level of detail required with a project level survey.

To improve visibility during the collection process, technicians switched from driving in a car to driving a golf cart equipped with the computer and microphone data collection system. The speed of data collection for project level information is about 2 to 3 miles per hour. After a data collection session, the technician downloads information into the host computer for processing and analysis.

The end result is a full Arcview database fully populated with spatially referenced data.

The project level data is captured at a very minute level. Cracks in the sidewalk, curb and gutter and the street are categorized by type and severity level. This data is used to determine the priority of repairs necessary at a given location. Information can be filtered based upon severity level and repair criteria. This allows for an accurate representation of the field information.

A custom work order system was bundled with the Vocarta software from Datria Systems to allow targeting specific areas for repairs. A work order is prepared for each repair area, which can then be included in a repair contract or given to street maintenance crews.

The speech recognition and mobile computer from Datria Systems’ Vocarta product has revolutionized the techniques used for infrastructure management. Not only are more data collected faster, but whole new categories of data can now be collected. Speech recognition and voice to data techniques led to a system that works faster and better.

Dave Chambers is the Manager of Street Services, City of Aurora, CO., 15740 E. 32nd Avenue Aurora, CO 80015303-361-2970, FAX: 303-360-0828 email: dchamber@ci.aurora.co.us.
Steve Bourgeois is vice president of product management and marketing for Datria Systems, 7211 S. Peoria St, Suite 260 Englewood, CO 80112, PHONE: 303-645-9310, FAX 303-645-9301, E-mail: steve.bourgeois@datria.com

SpeechTek Covers
for qualified subscribers
Subscribe Now Current Issue Past Issues