Acquire microsimulation model calibration & validation data using 1-hertz TLAP

Introduction: While this tool will not be the best choice for many situations, when conditions require what 1-second TLAP does best, it is highly cost-effective and it will enable your team to create and calibrate complex microsimulation models that might not otherwise be successfully completed. If you call Skycomp and describe your potential application, we can tell you about the strengths and weaknesses of all of your options, and we will refer you to an appropriate expert if an alternative method will provide you with better or more cost-effective results.

Discussion: Before investments can be made to improve mobility and eliminate bottlenecks, a process must be followed involving engineering analysis, stakeholder consensus building, and staged approvals. A well-built microsimulation model can be invaluable for all steps of this process. However, creating models of networks where travel patterns are complex pose two special challenges: first, how do you measure the complicated traffic flows that your model must replicate accurately? And second, if you have done that successfully and built a quality model, how do you establish its credibility with the audiences it was intended to enlighten (particularly when there are groups determined to fight anything you do)?

In theory, this is what surveillance drones could help us with in the future, after the legal and liability issues have been resolved that today preclude the use of high-altitude drones above populated areas. Cameras could record the movement of all vehicles, allowing the tracing of routes and true O-D patterns, or showing changing levels of congestion severity, or even parking occupancy and turnover rates, travel times, speeds & delays on any route, traffic signal operations, or simply enabling the counting of volumes or turning movement counts anywhere that is visible. The resulting variety of metrics and the granularity could not be matched by any other technology, at any cost. Next, aerial imagery could be juxtaposed with model calibrations to emphatically show that the model replicates reality. Lastly, additional metrics could be taken from the imagery later if additional needs are thought of.

Inspired by this vision and pressed by clients, Skycomp began in 2012 to develop the capability to provide exactly these services, but fully sanctioned by the FAA. Renting local, unmodified helicopters near each survey site was the only way to keep it affordable, which meant that a walk-up mounting system had to be invented for the camera clusters. Other computer-assisted processes then had to be assembled to align all imagery and then extract the many types of traffic flow metrics. Success followed such that today Skycomp deploys helicopters that hold fixed-position hovers one mile high, and are able to image all specified highway links in two-hour continuous sets, at one frame-per-second (1-hertz or 3,600 high-resolution digital images per camera per hour).

Through 2016, Skycomp has successfully conducted 80 TLAP surveys of this type, in 17 states. These have involved deploying from one to four helicopters running from one to 16 cameras simultaneously, recording traffic flows on up to 16 linear miles of highway. All of these surveys have enabled clients to receive TMC’s, volume counts, O-D patterns, travel times and/or other metrics, “pre-balanced” because they were all acquired simultaneously, to build complex models with ease and confidence that could not have been achieved with any other tool. (Interestingly, Skycomp has frequently been asked to go back into the imagery later to examine areas and extract metrics that were not previously thought of.) Furthermore, different metrics acquired from the TLAP imagery can often be used for the very different tasks of first calibrating the model (e.g. using ramp volumes, TMC’s and/or O-D patterns), and then validating it with different metrics (e.g. queue development & queue lengths, and/or travel times).

 

Stoughton Rd

 

 

Acquire model calibration data from INRIX (and optional hybrid survey w/ TLAP validation)

Introduction: The newest kid on the O-D block may be positioned to play a big role in traffic data collection in the near future. After Skycomp pilot-tested INRIX’s new Trips database product against a verified TLAP O-D dataset, Skycomp and INRIX agreed to join forces to promote and distribute this new offering. Skycomp now can take orders for INRIX Trips data directly (for data archived as far back as January 2014); Skycomp has developed the tools and expertise needed to dig into the raw INRIX databases on your behalf, to produce O-D and/or travel time profiles to fit your study needs. For borderline applications where there may be uncertainty deploying the INRIX option, Skycomp has been using a hybrid TLAP/INRIX methodology to test the accuracy of INRIX Trips and validate its use for specific study areas.

Discussion: While developing the 1-hertz TLAP service described above, Skycomp also kept an eye out for the arrival of other new technologies and tools. Skycomp was therefore intrigued to hear from INRIX that a new Trips database product was being readied for market deployment. This was to contain compilations of the millions of time-stamped “ping” records that INRIX receives every minute (for its real-time speed products) into individual historical vehicle trips, complete with start- and end-points and the intermediate waypoints. While this could potentially be invaluable for many types of transportation planning studies and model calibration needs, both macro- and (hopefully) micro-, questions were bound to arise as to its overall suitability for each application. How exactly could it be verified that the vehicles contained in the INRIX database were representative of the overall traffic stream (in my locality, on my roads)? If valid, how big would the sample sizes need to be to produce valid results? Would audiences accept models calibrated by this tool? And lastly, at what level of (micro-) granularity could it be used?

INRIX recognized that Skycomp, with its 1-hertz TLAP helicopter service, was uniquely capable of finding answers for these questions, and therefore could help clients gain confidence with this tool. “Hybrid” studies could be conducted that would use TLAP coverage to validate the INRIX outputs during pre-planned overlapped periods. Likewise, Skycomp recognized how the INRIX product was complementary to TLAP, because of TLAP’s spatial and temporal limitations. INRIX, with nothing to deploy or install and with an ability to reach back in time for archived data, would be virtually unlimited both spatially and temporally. The method would also benefit from cost efficiencies as coverage areas and time periods increased. TLAP could be deployed just for the critical heart of a large study area, while INRIX covered the rest, and for much longer periods. The granular TLAP would give users an appropriate level of confidence in the INRIX findings for the overlapped areas and periods (not to mention the other metrics contained in the imagery), and then the INRIX-only outputs could be used for the rest with an appropriate level of confidence.

To test these concepts, INRIX furnished data to Skycomp for a test O-D analysis against a TLAP dataset prepared for the Port Authority of NY and NJ, covering the New Jersey approaches to the George Washington Bridge. The TLAP / INRIX correlation was close enough that the two companies concluded a partnership agreement, whereby Skycomp would become an authorized distributor of INRIX “Trips” database product. Skycomp would conduct hybrid surveys based on client preferences, or would use its expertise working with the raw INRIX databases to produce INRIX-only output tables. (At the end of 2016, several more hybrid INRIX / TLAP analyses were nearing completion; please contact Skycomp for an update on the correlations found during the latest surveys.)

INRIX

 

Skycomp documents and quantifies traffic mobility or parking patterns and any underlying causes of congestion using images acquired during extended TLAP patrol flights (repeating tours that typically last from five minutes to one hour).

Introduction: Patrolling TLAP survey flights have been a Skycomp staple for 45 years, yet remain surprisingly relevant for the modern era. Roadside video cameras are not everywhere (and are not normally recorded where they are), and while digital sensors can tell a wonderful “big data” story 24/7/365, often they cannot communicate what is really going on. With today’s focus on public participation and non-technical stakeholder involvement in the planning process, and the need later (for accountability) to show that invested funds have achieved their objectives, extended TLAP patrols make as much sense today as before. As we have heard from dozens of clients over the years, a photo – that is, the right photo – really is worth many words.

Discussion: Skycomp uses fixed-wing aircraft to cover large highway networks, or helicopters for urban street grids, to acquire TLAP coverage of 100% of an assigned highway or street network. Fixed wing aircraft may also be used for TLAP of specific points of interest that are widely scattered, such as system-wide transit parking lots. The sample intervals typically range from five minutes to one hour. Post-flight, all photos are sorted by viewpoint, and then tightly aligned to facilitate the detection of patterns and/or the taking of counts & other metrics. Fast-moving airborne platforms allow for regional or system-wide coverage areas, and produce TLAP imagery to document queue lengths over time, the underlying causes, and the extent of any spillover to other streets. The applications are to understand and quantify delays for, say, planning major projects, evaluating stadium events, or, for public relations purposes, documenting traffic conditions before and while a major facility is closed for rehabilitation (as happened when the New Jersey DOT recently closed the northbound lanes of the Pulaski Skyway near Newark Airport).

 

Meadowlands

Acquire micro-trajectories for driver behavior research

This is a specialized research application for 1-hertz TLAP (see description under TLAP Model Calibration), with special attention paid to measuring accurate second-by-second speed and lane changes, following distances, and gap acceptances for making turns across traffic or merging into new lanes. While imaging, a premium is placed on maximizing camera verticality, using higher-resolution lens-camera combinations across smaller coverage areas, and applying vehicle position tags with precision. This allows micro-movements of vehicles to be calculated, for greater understanding of driver behaviors while negotiating, say, freeway weaving or ramp merging sections, unusual signalized intersections, alternative-design facilities such as diverging diamond interchanges or modern roundabouts, transit buses that stop repeatedly on city arterials, highway links with narrow lane configurations, or while traveling along heavily-used two-lane highways.

 

 FHWA