Enhancing Active Transportation and Demand Management (ATDM) with Advanced and Emerging Technologies and Data Sources

Chapter 3. Active Transportation and Demand Management Applications

This chapter discusses the specifics of the active transportation and demand management (ATDM) solutions and approaches and how they can be enhanced using the technologies and data sources described in chapter 2. Figure 12 depicts the challenge and tremendous opportunity that is faced by the implementers of ATDM solutions. Nearly all ATDM and/or advanced transportation management system (ATMS) deployments have the same basic functions, they:

1. Collect and monitor data from available data sources.
2. Assess system performance to identify what adverse or abnormal conditions to which and agency or system should respond.
3. Evaluate and recommend what actions to take. This is often done using response plans or Decision Support Systems (DSS).
4. Implement the dynamic actions which includes dissemination of this information via 511 systems, dynamic message signs, traveler information websites, etc.

Through the emergence of new data sources and technologies, the way ATDM will be performed is expected to improve. These transformational changes include:

• Monitor System – Today, data are pulled in from various traditional vehicle detection sensors to monitor traffic flows, speeds, and volumes. This can include inductive loops, magnetometers, radar sensors, video image detectors, etc. Also, certain probe data sources like HERE and INRIX are readily used for these same purposes. As it relates to traffic accidents and other traffic-impacting conditions like construction, computer aided dispatch (CAD) and planned lane closure (PLC) systems are used to provide this information. Also, closed circuit television (CCTV) cameras are used to gather certain information such as details related to a traffic incident.
  • The Future: The data that informs us about congestion levels, volumes, and traffic accidents will come from the vehicles themselves (e.g., connected vehicles) and/or other applications (i.e., the smartphone navigation application Waze). Systems and sensors will be connected via IoT allowing the gathering of similar information from a variety of sources never before used as part of ATDM and traffic management systems. Lastly, crowd-sourced data will be more prominently introduced. These data sources will be more accurate and timely and will enable ubiquitous coverage of all roadways, vehicles, and facilities.
• Assess System Performance – Today, sensors and other technologies are used to assess system performance. Basic analytics algorithms such as ones used to calculate travel times are put in place to determine congestion levels. CCTV cameras are used to monitor roadways and potentially identify where accidents have or are occurring. In some cases, modeling tools are used to help assess system performance both in near-real-time or for after analysis.
  • The Future: Through the use of cloud computing big data lakes,⁽¹⁴⁾ on-line analytics engines, artificial intelligence (AI), and machine learning (ML), the ability to assess system performance will be easier, more intelligent, and faster and be able to take into account much larger sets of information.
• Evaluate and Recommend Dynamic Actions – Today, various ATDM algorithms and DSSs are used by transportation agencies to recommend dynamic actions or response plan methods. Analytic and simulation models are used to evaluate the potential effectiveness of the recommended dynamic actions. Decision support engines and algorithms determine the ATDM responses such as the activation of queue warning signs, variable speed limit (VSL) signs, dynamic lane management, parking management systems, and dynamic shoulder use.
  • The Future: AI, ML, and deep learning (DL) may be used by transportation agencies as the basis for decision making. The key benefit here is that these systems are more intelligent; therefore, they can learn and adjust themselves over time. They can also better manage the vast amounts of information that will be available. Big Data, Cloud Computing, Data Analytics and IoT will be at the forefront of gathering, storing and better analyzing data to make decisions. Blockchain and other high-end data storage and mining mechanisms will be utilized to securely manage the volumes of information in a secure scalable manner.
• Implement Dynamic Action – Today, agencies implement actions such as placing messages on highway signs, enacting different traffic signal or metering plans, and notifying commuters and other agencies (e.g., police and fire) via traditional communication methods. Messages are placed on highway message signs and ATDM message signing (VSL signs, queue warning signs, lane management signs, etc.), and radio and television media and e-mail and text message systems are used, as are website and phone applications.
  • The Future: Information will be pushed automatically into vehicles, connected and autonomous. For example, the recommended speed limits messages that traditionally would be posted on overhead gantries would now be pushed into the vehicles via aftermarket devices (smartphone) or directly into the vehicle dashboard. Communication will occur with fleets of autonomous vehicles that are electric and shared (FAVES) as well as other mobility on demand services. More advanced smartphone applications will be available that enable all levels of crowd sourcing.

3.1 Active Transportation and Demand Management Approaches

This section discusses specifics on how new technologies and data sources can be used to improve ATDM solutions.

Active Traffic Management using Emerging Technologies and Data Sources

Active traffic management (ATM) deals with dynamically managing recurrent and non-recurrent congestion based on prevailing and predicted traffic conditions. It increases throughput and safety through the use of integrated systems and includes automation of dynamic deployment to optimize performance quickly without the delay that occurs when operators must deploy operational strategies manually. ATM approaches focus on influencing travel behavior with respect to lane/facility choices and operations.

The use cases and descriptions later in this section discuss operational ATM solutions and how they can be improved through the enhanced technology and data sources identified in chapter 2. It should be noted that many ATDM deployments include multiple solutions at once, i.e., a "typical" ATM corridor often employs dynamic lane use, dynamic shoulder use, VSLs and adaptive ramp metering simultaneously.

A clear example of how ATM is enhanced via an innovative new technology or data source is through the use of connected vehicle applications (e.g., vehicle to infrastructure [V2I] and vehicle to others [V2X]), which include smartphone V2X applications. Today, for most ATM applications, intrusive or non-intrusive traffic detectors are used to assess real-time traffic flows for ATM algorithms. Other incident detection methods are used to identify non-recurrent conditions such as crashes. These technologies include CAD systems, video image processing systems (VIPS), and automatic incident detection (AID) algorithms. Through the use of connected vehicle data, more granular traffic probe data can be obtained directly from vehicles. These data are much more granular and accurate; hence they can improve the timeliness and accuracy of many ATM functions. These data can augment or replace other current data sources used today. Also, dissemination of information to the motorist can be greatly improved through connected vehicle technology, including connected vehicle applications. For example, dynamic speed limits recommendation can be transmitted directly to the vehicle rather than via expensive roadside signs. Similarly, queue warning, weather warning, and dynamic shoulder use messages can go directly to the driver, which will improve overall operations and maintenance (O&M) costs. See the end of this section for descriptions of how advanced and merging technologies and data sources that can provide benefit to ATDM, mapped against the steps of the active management cycle.

Active Demand Management using Emerging Technologies and Data Sources

Active demand management (ADM) uses information and technology to dynamically manage demand, which includes redistributing travel to less congested times of day or routes or reducing overall vehicle trips by influencing a mode choice. ADM seeks to influence more fluid, daily travel choices to support more traditional, regular mode choice changes.

Some example ADM applications that can be improved via enhanced data sources and technologies include dynamic routing, pricing, and predictive travel times. Systems that have deployed predictive travel times include the San Diego Integrated Corridor Management System (ICMS), which leverages microsimulation tools and historical data to implement traffic predictions. Today, ML can be used to improve these same capabilities as can big data, cloud computing, and new probe-based data sources. Implementation of dynamic routing can be a tricky task, especially the methods that are used to recommend alternate routes. This is where the use of smartphone applications including crowdsourced applications may be beneficial. If a transportation agency desires to have traffic rerouted to avoid a serious accident, they can leverage these applications to direct drivers to the alternates that they prefer, such as alternatives where they can adjust traffic signal timings accordingly. Dynamic pricing algorithms can be improved through the use of more enriched data sources (e.g., connected vehicle data, probe data, etc.) where big data and cloud computing can be leveraged for the best dynamic pricing options/recommendations.

See the end of this section for descriptions of how advanced and merging technologies and data sources can provide benefit to ATDM, mapped against the steps of the active management cycle.

Active Parking Management using Emerging Technologies and Data Sources

Active parking management (APM) is the implementation of dynamic management of parking facilities to optimize performance and utilization of those facilities while influencing travel behavior at various stages along the trip making process: i.e., from origin to destination. Dynamically managing parking can affect travel demand by influencing trip timing choices, mode choice, and parking facility choice at the end of the trip. This ATDM approach can also have a positive impact on localized traffic flow by providing real-time parking information to users and ensuring the availability of spaces to reduce circling around parking facilities.

A key example where APM can make the best use of enhanced technologies and data sources is for dynamic wayfinding. Today, systems like the San Francisco Smart Parking and Los Angeles Express Park perform parking availability detection using traditional means such as streetline detectors or parking proximity sensors. Newer technologies can do this in a more efficient manner, e.g., CCTV cameras with video analytics can monitor large parking lot areas with a single camera.

Also, the means to direct people to available parking spaces can be done using smartphone applications in addition to roadside message signs.

Active Transportation and Demand Management Improvements using Emerging Technologies and Data Sources

See below for a depiction of advanced and merging technologies and data sources that can provide benefit to ATDM, mapped against the steps of the active management cycle.

Stage 1 — System Monitoring Enhancements for Active Transportation and Demand Management

Monitoring for system conditions such as roadway congestion levels, traffic incidents, construction activities, and weather conditions can be improved or enhanced by using these data sources or technologies:

• Light detection and ranging (LiDAR).
• Crowdsourced data, e.g., Waze.
• Automatic vehicle location (AVL) data.
• Global positioning system (GPS) probe data.
• Connected vehicle data.
• Mobile sensors/drones.
• Real-time turning movement counts.

Also, new mapping technologies such as crowdsourced mapping and high-definition (HD) mapping can assist with the common operating picture (COP) used within system monitoring tools. The use of cloud computing, big data, and data analytics can aid with performing system monitoring, especially as it relates to monitoring large amounts of complex data.

Stage 2 — Assess System Performance for Active Transportation and Demand Management

Within current ATM solutions, real-time systems utilize algorithms and other basic analytic tools to assess system performance. This can include real-time COP mapping for observing conditions as they unfold. There are some modern and emerging technologies that can improve how, and how accurately, performance is assessed. Some of the best emerging technologies include the use of big data and data analytics engines to assess the vast amounts of data that are being collected. In the future, there will likely be much larger and more granular data available from likely sources such as connected vehicle data. One of the key benefits of big data analytics is that they can aid at providing insights into data correlations that humans would not naturally see.

Other key technologies that can aid in assessing system performance are AI tools, which include ML and DL. Through programming with AI tools, the conclusions of system performance can be greatly improved. One example is that AI can aid with prediction of traffic conditions and events, which is a much more proactive approach, i.e., ATDM systems will then be able to assess not only current performance but future performance.

Performance can also be assessed in real-time using more advanced video analytics technology, which includes solutions that can identify traffic flow performance issues or other traffic event or incident issues. This includes wrong way driving, stalled vehicles, fire and smoke detection, etc.

Stage 3 — Evaluate and Recommend Dynamic Actions for Active Transportation and Demand Management

The next stage of the active management cycle is to evaluate and recommend dynamic actions based upon the performance assessments completed during Stage 2. This is where DSS can be improved. AL, ML, and DL can enhance DSS, which includes the concept of basing decisions on future predictions rather than just upon current conditions only. On-line microsimulation tools offer a way to assess and evaluate strategies in real-time to determine which strategies will work most successfully.

Cloud computing and more advanced data analytics can be used as well to evaluate the dynamic actions. With the appropriate computing power and/or cloud computing environments, these analytics can be done very quickly. Edge computing can also be used to analyze conditions and behavior such as video analytics, pedestrian detection, and poor local traffic signal performance.

Stage 4 — Implement Dynamic Actions for Active Transportation and Demand Management

New emerging technology can be used to implement the ATDM actions as well. This includes the following:

• Recommendations on how drivers should behave can be pushed into vehicles using connected vehicle applications or directly into vehicle dashboards that will be equipped in the future to receive and display these messages. These messages can include VSLs, queue warning, motorist warnings, weather warnings, and signal phasing and timing (SPaT).
• Edge computing is now being used to implement actions at the edge. For example, DMSs related to ATDM functions can be activated directly using edge computing rather than via centralized systems. Traffic signal adjustments and preemption can also be done at the edge. Moving computing and decision making to the edge can be more efficient, decrease costs, and speed system response.
• Narrow AI in the form of voice driven assistants can help with quickly implementing dynamic actions. Activation can occur with a single voice command that can replace 10+ mouse clicks on the system user interface.
• Smartphone applications and connected citizen applications can be used to implement alternate applications. For example, advanced parking applications can be used to notify motorists of available parking as well as arrange reservations and payment in advance. The same applications can be used to recommend mode shift and alternate routing.
• Feedback into crowdsourcing applications may also be useful. For example, informing the navigation crowdsourcing applications like Waze as related to blocked lanes or road performance deficiencies can assist with more effective ATDM implementations.

Use Cases

This section discusses use cases for applying emerging technologies and data sources to improve how ATDM is accomplished. The use cases provide examples of what is being used now, and what can be enhanced using new technologies and data sources.

Use Case 1: Recurring Congestion on Freeways and Highways

Brief Description

On weekday mornings, recurring freeway congestion occurs as commuters travel into the metropolitan area to report to work.

Actors

1. Motorist.
2. Transportation management center (TMC) operator.
3. Traffic engineer.

Pre-conditions

1. ATMS software is running in the TMC, and a TMC operator is able to take actions to reduce the congestion.

Basic Flow: Currently in Use

Figure 13 is a representation on this use case without advanced or emerging technologies and data sources applied.

A description of the stages follows:

1. Traditionally, vehicle detection devices such as inductive loop sensors, radar sensors, or video analytic sensors detect congestion buildup on the freeway and alert the TMC operator of the congestion build up on an ATMS map. Cameras along the roadway can also be used to confirm the congestion.
2. The TMC operator enters or confirms a congestion event (the existence of excessive congestion) into the ATMS and generates the recommended actions to implement.
3. System actions are recommended to help deter or reduce the congestion levels or minimize its impact.
4. Several potential strategies are implemented to address congestion:
   1. Activate or adjust ramp metering rates.
   2. Implement junction control.
   3. Implement dynamic pricing.
   4. Implement part time shoulder use.
   5. Implement dynamic speed limits.
   6. Implement alternate routing.
5. The TMC operator monitors the impact of the strategies that are deployed.
6. Data are stored in a historical database. Traditional reports can be used to determine performance metrics.

Basic Flow: Emerging Technologies and Data Sources

Figure 14 is a representation on this use case with advanced or emerging technologies and data sources applied.

A description of the stages follows:

1. Traditionally, vehicle detection devices such as inductive loop sensors, radar sensors, or video analytic sensors detect congestion buildup on the freeway and alert the TMC operator of the congestion build up on an ATMS map. Cameras along the roadway can also be used to confirm the congestion.
2. New data sources are also used to assess congestion levels, including:
   1. Crowdsourced data (e.g., Waze).
   2. Connected and autonomous vehicle data.
   3. Real-time turning movement count data.
   4. Data analytics from cloud computing.
3. The TMC operator enters or confirms a congestion event (the existence of excessive congestion) into the ATMS and generates a traditional logic-based response plan.
4. The response plan is recommended to help deter or reduce the congestion levels or minimize its impact. Traditional DSS methods are used, as are AI, ML and DL methods. Data Analytics and on-line microsimulation helps to help assess and evaluate the recommended action plan.
5. Several potential strategies are implemented to address congestion:
   1. Activate or adjust ramp metering rates.
   2. Implement junction control.
   3. Implement congestion pricing.
   4. Implement dynamic shoulder use.
   5. Implement dynamic speed limits.
   6. Implement dynamic routing.
   7. Send recommendations using smartphone applications.
   8. Send connected vehicle messages to smartphones and aftermarket devices.
6. The TMC operator monitors the impact of the strategies that are deployed.
7. Data are stored in a historical database. Traditional reports and cloud-based data analytics methods are used to determine performance metrics.
8. Blockchain data storage can also be used, especially as it relates to more sensitive data and data that warrant more secure storage.

Post Conditions

Successful Condition

Congestion is detected more quickly.

Congestion and delay levels are more accurate.

More appropriate implementation actions are recommended and implemented.

The system learns and improves itself over time.

Better analytics and reporting.

Recurring freeway congestion is measurably reduced.

Failure Condition

Detection times and congestion level accuracy are only marginally improved.

Freeway congestion is not reduced.

Implementation improvements are only marginal when compared to cost.

Use Case 2: Major Incident on Freeway

Brief Description

A typical ATDM use case is related to how a major freeway incident is detected, the decisions made on how to respond to the incident, and the method(s) to respond to the incident. Figure 15 represents a use case on how a typical freeway incident is handled, keeping in mind that there are different options and variations that can be employed.

Actors

1. Motorist.
2. TMC operator.
3. Traffic engineer.

Preconditions

1. ATMS software is running in the TMC, and a TMC operator can take actions that are used manage the incident. The following actions are available for incident response:
   1. Post a 511 message.
   2. Post a message on a sign.
   3. Dispatch tow trucks/service patrol vehicles.
   4. Dispatch fire and/or police.
   5. Implement ramp metering.
   6. Adjust signal timing.

Basic FLow: Currently in Use

1. Traditionally, a potential incident is identified by the following mechanisms:
   1. A traveling motorist calls 911 to report the incident.
   2. Data are received from law enforcement agencies' computer aided dispatch (CAD) system.
   3. Video analytic sensors detect congestion buildup.
   4. Vehicle detection devices such as inductive loop sensors and radar sensors detect congestion buildup on the freeway and alert the TMC operator of a potential incident using an incident detection algorithm.
2. The TMC operator confirms the incident.
3. The TMC operator generates a traditional rules-based response plan.
4. The response plan is implemented by the TMC operator:
   1. Post a 511 message.
   2. Post a message on a sign.
   3. Dispatch tow trucks.
   4. Dispatch fire/police.
   5. Implement ramp metering.
   6. Adjust signal timing.
5. Details regarding the event and its response plan are archived into the database.
6. Traffic engineers can generate traditional reports that can be used to determine performance metrics.

Figure 16 characterizes the same freeway incident management use case but introduces new data sources and technologies.

Basic Flow with Emerging Technologies and Data Sources

1. A potential incident is identified by the following mechanisms:
   • A traveling motorist calls 911 to report the incident.
   • Data are received from law enforcement agencies' CAD system.
   • Video analytic sensors detect congestion buildup.
   • Vehicle detection devices such as inductive loop sensors and radar sensors detect congestion buildup on the freeway and alert the TMC operator of a potential incident using an incident detection algorithm.
   • ML is used to predict incidents.
   • Waze crowdsourcing is used to identify incidents.
2. The TMC operator confirms the incident:
   • AI auto-confirms the incident.
3. The TMC operator generates a traditional rules-based response plan:
   • Big data in the cloud computing environment, along with artificial intelligence are used to develop a response based on historical and predictive data.
4. The response plan is implemented by the TMC operator:
   • AI auto-implements the incident response:
     1. Post a 511 message.
     2. Post a message on a sign.
     3. Dispatch tow trucks.
     4. Dispatch fire/police.
     5. Implement ramp metering.
     6. Adjust signal timing.
5. Details regarding the event and its response plan are archived into the database:
   • Data are archived into the Cloud where they are available for input into predicting incidents or generating response plans.
6. Traffic engineers can generate traditional reports that can be used to determine performance metrics:
   • Data analytics transform and model data to support decision making.

Post Conditions

Successful Condition

The incident is cleared, and traffic is cleared in a timely manner.

Failure Condition

The incident is cleared, but not in a timely manner, or causes additional incidents.

Use Case 3: Parking Near the Airport

Brief Description

In preparation for a week-long trip, a traveler is examining options for parking his vehicle near the airport.

Actors

1. Traveler.
2. Parking system operator.

Preconditions

1. Parking facilities exist near the airport.

Basic Flow: Currently in Use

Figure 17 depicts a typical airport parking use case.

1. Sensor systems are constantly monitoring the spot usage and availability using traditional sensors.
2. In advance, the traveler manually selects a parking provider that provides the ability to reserve a parking spot in advance. This reservation may or may not be at the closest facility to the airport. The fees are generally based on a fixed schedule; however, some facilities have currently implemented dynamic pricing.
3. Alternately, the driver proceeds to the airport and pays fees directly at parking structure.
4. Based upon parking availability and pre-existing reservations, the system provides direction to the driver, e.g., proceed to this level, aisle, and spot.
5. If parking is full, driver is directed to another off-site parking option.
6. Depending on demand, dynamic priced parking may be activated.
7. The traveler then parks his vehicle in the reserved spot.
8. Data are stored for further analysis by the parking system operator.

Basic Flow: Emerging Technologies and Data Sources

Figure 18 represents the same parking system use case but uses merging technologies and data sources.

1. Sensor systems are constantly monitoring the spot usage and availability using traditional sensors.
2. New technologies and data sensor types are used to monitor system parking availability, including video analytics sensors and LiDAR.
3. In advance, the traveler manually selects a parking provider that provides the ability to reserve a parking spot in advance. This reservation may or may not be at the closest facility to the airport. The fees are generally based on a fixed schedule; however, some facilities have currently implemented dynamic pricing.
4. Alternately, the driver proceeds to the airport and pays fees directly at parking structure.
5. Based upon parking availability and pre-existing reservations, the system provides direction to the driver, e.g., proceed to this level, aisle, and spot.
6. AI is used to automate all parking system actions.
7. Cloud computing and data analytics are used to predict parking availability and adjust dynamic priced parking to maximize revenue.
8. If parking is full, driver is directed to another off-site parking option.
9. Depending on demand, dynamic priced parking may be activated.
10. The traveler then parks his vehicle in the reserved spot.
11. Data are stored for further analysis by the parking system operator.
12. Secure data such as transactions and license plate reads are stored using block chain.
13. Data are analyzed using data analytics and cloud computing.

Post Conditions

Successful Condition

The vehicle is able to park in the reserved spot.

Failure Condition

The reserved spot was not available, leaving the driver to search for another.