Skip to content

Revolutionizing Transportation with GenAI’s Intelligent Systems

Introduction: GenAI pillars

The roots of GenAI can be traced back to the inception of the idea of Artificial Intelligence (AI), defined as programs capable of sensing, reasoning, adapting, and acting. It represents a comprehensive framework encompassing both theoretical and practical development of computer systems to perform tasks that typically require human cognition and intelligence.

We might consider that the beginnings of artificial intelligence itself date back to the mid-20th century, primarily due to the British logician and computer innovator, Alan Mathison Turing. Turing’s conceptualized computer revolved around a stored-program concept, where the machine could modify or enhance its own program. Turing frequently discussed how computers could learn from experience and solve new problems using guiding principles—a method now known as heuristic problem-solving.

Another fundamental component crucial to today’s GenAI is Machine Learning (ML), emphasizing the “adapt” aspect of AI through algorithms that bolster program performance by ingesting and learning from increasing amounts of data over time, without needing explicit programming.

The third notion that has been significantly advanced within Machine Learning algorithms is Deep Learning, wherein Multilayered Deep Neural Networks autonomously learn from extensive data, adapting through each iteration. On its own, Deep Learning has transitioned from initially being a discriminative model to a Generative Model, where networks learn from the probability distribution of input data, enabling them to generate new data instances based on learned data logic.

Finally, we encounter Generative AI (GenAI), which employs a subset of Deep Learning methods to create new content post extensive data assimilation and prolonged training phases. The primary distinction between GenAI and conventional AI is GenAI’s capability to function without labeled examples. Instead, GenAI detects patterns and structures within data, leveraging this comprehension to generate entirely novel content, ranging from textual responses and artwork to musical compositions and realistic human voices.

Applying GenAI in motorway management

The potential of GenAI is limitless, and we are only beginning our exploration. A new era has dawned, presenting significant prospects for drastically improving our professional and personal lives across various scenarios.

Numerous sectors and topics can be discussed to explore GenAI’s applications. However, for this discussion, I will concentrate on its application in the sector I’ve been involved with for over 18 years: Interurban roads. Below, I have outlined a brief list of potential use cases for GenAI in interurban roads, emphasizing the transformative advantages it can provide.

  1. Video analysis: incident detection

Within Intelligent Transportation Systems (ITS), commonly implemented on motorways (such as SOS phones, CCTV cameras, variable message signs, and vehicle detection systems), it is also conventional to deploy Automatic Incident Detection (AID) systems based on cameras, usually placed at critical infrastructures or certain road segments like tunnels. These systems have been employed for more than 20 years, incorporating technological advancements over time. Not only by utilizing better cameras but also by integrating AI and ML elements into incident detection algorithms to minimize false alerts caused by phenomena like light reflections, birds, or shadows. Despite these enhancements, many incident detections remain limited, with a non-negligible number of false positives that still require operator verification before action can be taken.

GenAI introduces the concept of Large Vision Models (LVM), referring to sophisticated AI models designed to process and comprehend visual data, such as images or videos. These models are “large” due to their considerable number of parameters, often reaching millions or billions, enabling them to learn intricate patterns in visual data.

One notable characteristic of AID systems is the necessity to program each use case individually, addressing numerous typical, frequent, and critical scenarios. Each is incorporated into the system in a specific way, with prescribed parameters and constraints, leaving many other potential use cases unaddressed. This means several possible scenarios that could be critical or significant are entirely overlooked by the AID system as they don’t conform to any of the limited programmed use cases.

Applying GenAI algorithms to the same video footage, the system will be able to understand the video as a whole, recognizing all types of anomalies without fitting them into a predefined set of scenarios. Instead, it assesses from a broader perspective, similar to an operator, identifying elements that stand out in a video for various reasons, e.g., detecting non-vehicle objects on the motorway (pedestrians, animals, lost cargo); identifying dangerous driving behaviors: sudden lane changes or abrupt speed shifts. It can also efficiently distinguish between birds, shadows, or reflections, thus reducing a broader range of false positives.

Upon detecting an anomaly, the GenAI system can classify it by severity and take necessary actions automatically or generate a comprehensive report in an optimal format, incorporating all essential information for easier inspection and understanding by an operator. This report could include, for instance, the number of involved vehicles, vehicle brands, models, colors, the number of people in the scene, whether children or elderly individuals are present, the involvement of police or fire brigade personnel, road conditions such as water or oil spills, and this can be provided in written form or supplemented with sketches, images, or video creations to facilitate the operator’s understanding.

It is significant to emphasize the term “understand”: While an AID system may detect several isolated events: Smoke, slow vehicle, lane change, vehicle driving on the shoulder, or stopped vehicle; the GenAI system can comprehend the video scene as a whole, correlating all incidents into a single action from start to end, potentially summarizing the previously mentioned sequence of events as “vehicle breakdown” with a holistic approach. This capability for GenAI is achievable through the accumulation of vast amounts of data, discerning what constitutes “normal” and incorporating additional information from diverse sources, providing GenAI with a broader perspective and deeper understanding of our environment. This enables a higher comprehension level when analyzing a scene and producing a highly detailed report in the necessary format to almost ensure the report’s clarity.

  1. Video analysis: Prevention

Applying identical GenAI algorithms to large volumes of historical and real-time videos,

GenAI will be a pivotal innovation for alleviating or reducing congestion, ensuring seamless traffic flow or accidents prevention, and improving road security. The GenAI algorithms will possess the capability to discern patterns that cause congestion or accidents well before they occur, enabling real-time interventions like adjusting speed limits or issuing pertinent messages via variable message signs or fine-tuning traffic light timings in ramp metering systems that regulate the input and output flow of traffic onto the highway.

GenAI will have a comprehensive perspective and comprehension of the entire highway and its environment, allowing for smarter decisions as an integrated system that functions holistically, not just in isolated locations, accounting for video feed data (historical or real-time) alongside large volumes of additional data influencing its decisions to optimize outcomes such as the season, the time of day, weather conditions, types of vehicles in the vicinity (cars, trucks, emergency, etc.), current traffic conditions, and anticipated traffic trends on current or nearby roads or towns and much more.

Additional types of systems that GenAI might be able to interact with, if present on the highway, would include delivering information messages directly to users through radio, SMS, mobile apps, or even straight to connected and autonomous vehicles via specific roadside equipment, providing them with traffic information and recommendations on speed, lane selection, best route alternatives, etc. It could also send alerts to users before embarking on their journeys about the existing situation and the best available options: Routes, public transportation, best times to travel, etc. all with a significantly higher level of understanding and information conveyed compared to what current systems offer.

  1. Video Analysis: Simulation

Another potent tool that GenAI will greatly enhance is simulation. Via a written or verbal chatbot, the operator can describe a specific traffic scenario and inquire about the optimal approach to manage the situation, whether to prevent it, optimize it, minimize its impact, increase safety, or all together. The GenAI algorithm will be capable of presenting a list of options with their advantages and disadvantages, considerations, and even a recommended choice. Once more, as previously mentioned in other cases, the fact that GenAI will be considering enormous amounts of historical information and numerous extra data and influencing factors will equip operators with the best possible execution plan for a given situation.

For instance, types of situations that might be valuable for analysis include:

  • Origin – destination route for transporting hazardous materials or large cargo
    System output: Required documentation, ideal date and time, speed, specific route, transportation and road signaling, measures in case of an accident depending on the goods being transported, etc.
  • Planning for specific days with high traffic expectations: Festivities, holidays, concerts, etc.
    System output: Road messaging and signage at different dates and times, recommended speed limits, necessary traffic management: where and how, etc.
  • Planning maintenance works: lane closures, deviations, speed limits, etc.
    System output: Road messaging and signage at different dates and times, best working area signaling, workers’ safety measures, work efficiencies to work smarter and faster, etc.

In all scenarios above, it would be required but quite straightforward to incorporate possible specifics such as the existence of work zones, occurrence of accidents, varying levels of heavy vehicles, etc. to be considered by the GenAI as part of each scenario.

  1. Maintenance and Technical Support

In this context, the application is not exclusive to roads and can be applied across all types of maintenance and technical support. With GenAI, maintenance personnel or technicians will no longer need to rely on their organizational skills, recall which document they need for a specific job, etc. GenAI makes it very simple to implement a basic chatbot, usable either in writing or verbally, where workers can have a conversation in any language, with the interface where the GenAI engine will be able to explore vast amounts of unstructured data providing the required information, from the right document, on the correct page, within a fraction of a second!

The same kind of chatbot can also function as a user interface for technical support, where the GenAI algorithms can facilitate a comprehensive dialogue, aiding in resolving the problem using internal documentation and even supplementing it with information from the Internet. Furthermore, it can produce a highly detailed ticket for the technicians to generate custom schematics, drawings, representations, images, etc., pre-verified with the user and appended together with a thorough description, a table with all user and device or devices information for a clearer grasp of the problem, and even a list of necessary tools to solve the issue.

Overall, the integration of GenAI algorithms into our lives will bring substantial improvements, optimizing our professional and personal experiences through its ability to analyze and process millions of data from thousands of diverse sources. Specifically in the interurban roads sector, various use cases will not only aid operators and maintenance teams in enhancing their work but will introduce automatic real-time analytics and responses to numerous real-time situations at a level no human could achieve, with more precise and well-studied outputs that will contribute to safer roads, improved traffic flow, optimized journeys, accident prevention, and potentially many other advancements in the near future.

GenAI has just emerged and is here to stay, transforming our world fundamentally. However, it comes with a dual aspect; while it will significantly enhance our lives, it will also introduce new scams, identity theft, and many other threats we never imagined or had to worry about previously. Thus, it is crucial to harness the benefits of the new GenAI algorithms but even more vital to learn and be aware of the new dangers we will face, which could make us much more vulnerable.

Leave a Reply