Anatomy of a mobile mapping system

September 5th, 2014, Published in Articles: PositionIT, Featured: PositionIT

 

Mobile mapping systems are ground truthing soldiers of accuracy and produce high-quality data to mapping data providers. Popularised by Google’s Street View cars, various mapping data companies operate similar vehicles to collect an array of data for various uses.

International mapping data provider, TomTom, is one such company, and recently upgraded their older mobile mapping panel door vans to tougher vehicles with newer technologies and sensors to survey more terrains and collect higher
quality data.

Fig. 1: The tablet from which the sensors are monitored and the surveying software operated, connected via a remote desktop connection to the server where the actual processing of data takes place.

Fig. 1: The tablet from which the sensors are monitored and the surveying software operated, connected via a remote desktop connection to the server where the actual processing of data takes place.

The new high-clearance vehicles, Ford Ranger 2x4s, will collect mapping data by driving road networks across South Africa as well as other parts of southern Africa, and can survey along varying road surfaces and unpaved roads when required. Currently, the collected data is only used in-house to develop and improve the company’s existing mapping information and products.

The new vehicles, which are fitted with a ruggedised server, tablet computer, and nine sensors, were designed with input from the mobile surveyors who operate them. These sensors are fitted to a customised roof rack and metal arm at the back of the vehicle, and feed the collected data to the server fitted behind the passenger seat inside the vehicle.

The server with its dual hard drives (each 4 TB) is secured to the vehicle with special rubbers for shock resistance, and is where the calculations and storage of data takes place. It is in turn connected to a tablet computer via an Ethernet cable and a remote desktop connection. The server runs of the vehicle’s battery and therefore works only when the vehicle is on, and has to be shut down before the vehicle is turned off.

Fig. 2: One of two DGPS antennas connected to the NovAtel ProPak-V3 receiver inside the vehicle.

Fig. 2: One of two DGPS antennas connected to the NovAtel ProPak-V3 receiver inside the vehicle.

The tablet serves as an interface to the server, and is mounted on a customised metal stand on the passenger side, facing the operator. From here the surveyor sets up, controls and monitors the sensors, runs the surveying software, and controls the server.

Sensors and data

Besides using a more compact and ruggedised server, the new vehicles are also fitted with updated sensor technologies. These include a panoramic camera, three laser scanners (one of which is a high definition scanner), and a Differential GPS (DGPS) sensors/receiver. Inclination is measured by the inertial measurement unit (IMU).

The two DGPS antennae are mounted on the left front and back corners of the roof rack, and are integrated with a NovAtel ProPak-V3 unit receiver which receives a real-time differentially corrected GPS signal.

Fig. 3: The SICK laser scanner at the back of the vehicle.

Fig. 3: The SICK laser scanner at the back of the vehicle.

There are also two SICK LMS 511 lidar scanners, which scan 190 points per scan, with a frequency of 100 Hz and up to five echoes. One of the scanners is mounted on the right back corner of the rack facing upwards, and is used for collecting “corridor” data.

The second SICK lidar scanner at the back faces downward and collects road lines, traffic island and related data in a point cloud. These side-scan lasers generate point clouds for single-click measurement with panorama images, geo-positioned 3D representation of surroundings, as well as road surface markings

A collapsible short mast in the centre of the rack is fitted with a high definition Velodyne HDL-32E lidar scanner, with a Pointgrey LadyBug5 panoramic camera right above it. The USB 3 panoramic camera with its six 5 MP cameras, combined with synchronised software, delivers 30 MP high-resolution, 360° digital spherical panorama images.

Fig. 4: The two SICK laser scanners, one facing upwards and the other downwards, collect corridor data.

Fig. 4: The two SICK laser scanners, one facing upwards and the other downwards, collect corridor data.

Below the camera is a Velodyne 3D lidar scanner, which collects 700 000 points per second, and replaces five lidar scanners on the previous vehicles. The scanner captures a higher density 3D point cloud of buildings and surroundings, and allows faster driving speeds.

A Memsic IMU 800 inertial measurement unit, situated in the back of the vehicle, measures inclination. There is also a motion sensor mounted just above one of the brake drums on the back wheels.

The DGPS, IMU, and brake disk motion sensors are combined to record a highly accurate GPS track and geolocation of imagery. The lidar and imagery are also combined, to provide a point cloud with an image overlay. The complete system produces data with sub 10 cm accuracy.

The surveying process

After the sensors are setup and cleaned, the vehicle is switched on, followed by the server and tablet. To ensure accuracy, the motion sensor on the disk brake is calibrated once a
week or more if needs be. The surveyor uses the touch screen tablet to establish a remote desktop connection to the server, and then opens the Command Centre, Storage Control and Lady Bug image control.

Fig. 5: A side view of the mobile surveying vehicle.

Fig. 5: A side view of the mobile surveying vehicle.

This displays the system status and all the sensor feeds on the screen, allowing the surveyor to monitor the DGPS signal strength and image quality at regular intervals during surveying. (When insects and dust end up on the lenses, the surveyor has to stop to clean them.)

Once the system feeds and software are running, the surveying program itself, DAMP Day, is opened and the work log/routes are loaded. DAMP Day also indicates the optimal driving speed based on the quality of light, to ensure high quality data capture. Driving faster than the required speed, for example, diminishes the point cloud- and image quality, and can also lead to underexposed or blurred images. The survey is then initiated on the tablet display, and the data recorded as the surveyor drives, and saved to the hard drives in parallel to ensure redundancy.

Processing the data

Upon completion the recording process is stopped on the tablet, and the data on the server is reviewed for quality. Reviewing is done using OTM software, which draws up a random selection of images and data from the day’s surveying, which the surveyor clicks through and checks for quality.

Fig. 6: An overview of the layout of sensors in relation to each other on the custom roof rack of the surveying vehicle.

Fig. 6: An overview of the layout of sensors in relation to each other on the custom roof rack of the surveying vehicle.

The route map can be customised to account for routes under construction and other restrictions, which flag the roads to be covered in the next survey of the area. Once the quality check is completed, algorithms merge and compress the collected data, and embed the job sheet and metadata file.

When the hard drives are near full, they are removed and replaced with a new pair of drives. One of the removed disks is stored at TomTom South Africa’s offices, and the other shipped to Poland, where TomTom’s technical team extracts, processes and packages the data for the company’s different products.

Uses of the data

The two main mobile surveying data outputs are the Panoramix BEM (Bird’s Eye Mosaic) View and Panoramix mobile mapping imagery (same as Street View). The Panoramix BEM mobile mapping imagery is mosaicked and georeferenced to produce 2D representation of the road surface so as to accurately define lane information/geometry, locate manoeuvres such as turn restrictions, indicate painted road markings, and sign post position and attribution among other things.

Due to privacy concerns, the collected imagery is only used internally. Various applications are available internally for TomTom production teams to view the imagery and derive outputs to accurately update map datasets. Teams working on TomTom datasets are able to view imagery as located in reality, mirrored and synchronised in relation to the relevant TomTom street network database. The datasets are correlated so pans through the database will automatically skip through the mobile collected data as well.

The processed data is then integrated into the company’s existing data sets and used to check accuracy, refine navigation and do other checks. This data can be used in Advanced Driver Assistance Systems (ADAS) and for 3D asset management, and is available in TomTom’s Source Explorer, Cartopia, and OASIS applications.

Conclusion

TomTom operates two of the high-clearance vehicles described above, which are the company’s flagship mobile data collectors. They are operated in conjunction with the scaled down systems, which focus more on video recording.

The two main mobile surveying vehicles are assigned to survey the southern African region, starting with Namibia and Botswana. The surveys will start with the main road networks, before focusing on high-density areas such as major cities, and then zoning in on smaller roads to complete the data.

Mobile surveying data collected in this way has a high refresh rate, depending on the area in question. High density and areas of rapid development such as cities have to be surveyed annually, while lower density areas are surveyed at least once every two years.

Ever improving mobile scanning technology increases the quality, speed and functionality of mobile surveying and the collected data, making mobile surveying tools powerful data collectors of rich mapping data.

Send your comments to positionit@ee.co.za

Related Articles

  • AI accelerates local drones-for-social good solutions
  • Vineyard map opens doors to global regions
  • Developing an integrated data service ecosystem
  • Lidar enables Game of Thrones visual effects
  • Report: Extended reality needs responsible design