1-2-5
Figure 4: The Navigation Module
After the processing of the GPS/INS data the actual survey
trajectory along with accuracy estimates are stored in a
georeferencing file. The planning module then uses the
georeferencing file to highlight those segments which have
been surveyed using map matching techniques such as
slope-intercept and nearest future search methods, for more
details see Bullock, 1995.
3.3 The Expert Knowledge in the Quality Control
Quality control usually has a real-time and a post-mission
component. In the real time component, expert knowledge
decide whether a specific set of data is sufficient to provide
the required accuracy with a certain level of probability. In
the post mission component, knowledge decide analyze the
result and performance achieved in different environment
to increase the knowledge base of the system. If the results
of the real-time prediction differ considerably from the
results of post-mission analysis, the real-time model needs
improvement. This can only come from the analysis of
large discrepancies between prediction and post-mission
results. Thus, each real-time model includes a certain
amount of expert knowledge that has been gained in post
mission analysis. It is the art of real-time quality control to
combine this expert knowledge with the minimum
information on the measurement process and still to arrive
at reliable predictions. Such predictions would provide a
systematic operation procedure that guaranty the quality of
the results required for the mission.
Real-time Quality Control: The real-time quality control
component gives alerts to the operator in situations where
the specified trajectory quality can most likely not be
achieved.. These alerts are determined by the computer
which monitors data accuracy of GPS and INS, tracks
signal blockage for each satellite, and introduces expert
knowledge on INS bridging (smoothing) and the
convergence time of OTF ambiguity resolution techniques
into the process. In those cases, where an alert is given, the
vehicle is stopped to either allow an additional ZUPT or, in
more critical cases, an independent ambiguity resolution.
Using these additional measurements periods, the required
post-mission trajectory accuracy can be obtained in the
specified number of cases and re-surveys can be avoided.
Examples of expert knowledge rules used in the real-time
quality are:
1. Switching to the INS stand-alone mode of operation is
the number of tracked satellites is less than 4.
2. Alerting the survey crew to the start/end of the ZUPTs
by monitoring the time since the start of INS stand
alone mode of operations.
3. Monitoring the INS gyro rates during ZUPTs such that
they do not exceeds certain threshold, In those cases
where the gyro rates exceeds the threshold, due to
passing cars or movement inside the van, an alert is
issued to extend the ZUPT time. See Figure 5, for an
example of ZUPT splitting due to movement during
ZUPT.
4. Monitoring the GPS data such that if at least 4
satellites have been continuously tracked for certain
time period, defined in the expert system database as
the OTF convergence time, the expert system switch to
GSP/INS navigation mode.
5. Switch off the cameras if the van is outside the
planned survey area.
Post Mission Quality Control: In post-mission quality
control, all available resources are used to obtain the best
possible trajectory. This includes a suite of programs for
cycle slip detection and correction, INS bridging and
backward smoothing in case of GPS outages. The major
task of this segment is quality assurance, i.e. the certainty
that a specified percentage of the stored INS and GPS data
are of sufficient quality to allow a continuous computation
of the vehicle trajectory within the specified accuracy
limits. For example, the requirement to achieve a standard
deviation of 30cm in position for 98% of the post-mission
trajectory computation would be such a specification. The
post mission component should therefore provide all the
tools to provide necessary information to the expert
knowledge database. This information includes:
1. Satellite availability, start/end of GPS static mode and
of INS ZUPTs
2. OTF convergence time
3. INS bridging accuracy for both LI and wide-lane
4. Difference between INS and GPS solution