adimas - Fotolia
Fully autonomous vehicles are not yet a reality, but the future of AI in transportation is already arriving in other areas. Machine learning and cognitive technologies are being applied across many different corners of the transportation industry, from making our non-autonomous vehicles smarter to handling the wide range of shipping responsibilities.
Making cars smarter
According to automotive trade journal Ward's Auto, there are over 1.2 billion vehicles on the road today, with the total likely to cross over 2 billion by 2035. With all these vehicles on the road, it's no surprise that the number of fatal accidents is high. According to the World Health Organization, over 1.25 million people die each year as a result of traffic accidents. Automotive manufacturers, technology companies, and regulatory bodies are looking to AI and machine learning to help change that statistic.
But the future of AI in transportation will see advanced technology making driving safer. Rear-view cameras, blind spot monitoring, adaptive cruise control, lane keeping technology, and collision avoidance are allowing modern vehicles to be increasingly aware of the roads and surrounding vehicles. Machine learning plays a crucial role in making sense of all that data and delivering predictive warnings to the driver. Advanced Driver Assistance Systems use AI to learn the behavior of drivers and anticipate when problems might occur.
Additionally, cars are becoming more connected with each other. The vision of the connected car will enable communication between vehicles to create safer driving conditions and enhanced collision avoidance by alerting drivers to traffic jams miles ahead. Through machine learning and AI, vehicle systems learn about larger traffic patterns, even if the car isn't moving. Some cars are now being outfitted with internal cameras that monitor the driver. Using AI-powered facial recognition technology, these systems can alert the driver or even control the vehicle if it detects that the driver is impaired, drowsy, or otherwise distracted.
AI is also being used to improve the experience inside the vehicle. Voice assistants are making their way into the vehicle environment, giving drivers and passengers the ability to interact with features inside the vehicle such as climate control or the entertainment system without having to use their hands or take their eyes off the road. Vehicles are also making use of AI to help predict travel time to destinations, give suggestions for food or nearby points of interest on road trips, and automatically adjust climate and other settings based on learned driver and passenger preferences.
AI steering intelligent shipping
While the seas might not be as busy as the rush hour highway, shipping is still fraught with danger and difficulties and AI is being applied to make shipping more reliable, safer and efficient. The vast majority of goods purchased overseas are delivered via shipping containers on large vessels. Machine learning is helping to optimize delivery routes and improve logistics, saving both time and money. Predictive analytics is improving the accuracy of predictions for estimated times of arrival for container ships as well as identifying potential issues in shipping lanes and at ports.
Shipping companies are also using machine learning to analyze historical shipping data to find patterns among shipping seasons, products shipped and activity at ports. This allows shipping and logistics firms to optimize deliveries and the aggregation of different shipments across different shipping channels. Container terminal operations are also seeing efficiencies through predictive maintenance, supply chain optimization and handling shipping and documentation processes.
In the same way that AI technologies are being used to improve the human-machine interface in vehicles, so too are natural language interfaces finding their way into ship control bridges and control systems. Ship user interfaces are enabled with voice control, and cameras in crew areas are keeping an eye on the ship, its officers, and potential external threats. These camera systems use computer vision to watch out for criminal and piracy activity, providing continuous surveillance.
Like their better-publicized peers in the automotive industry, ships and vessels of all sorts are becoming enabled for autonomous piloting and operation. Rolls-Royce recently partnered with SVAN to demonstrate the first fully autonomous ferry, capable of docking, navigating and avoiding obstacles without any human intervention. Autonomous shipping can significantly reduce the costs of operating regular, repeated navigation routes. In fact, the autonomous shipping industry is predicted to be over $136 billion by 2030, with tugboats, small ferries and shipping container vessels being the first to be enabled with partial or full autonomous capabilities. Major shipping industry companies including Maersk, Panalpina, Flexport, CMA CGM and others are taking significant steps to advance the state of autonomous shipping in the industry.
Autonomous and intelligent trains are already here
Trains have long been enabled with intelligent technologies, with intelligent automation implemented over 50 years ago on the London Underground. The International Association of Public Transport classifies five Grades of Automation (GoA), ranging from a human being in full control of the train to completely unattended train automation. Surprisingly, there are many fully autonomous GoA Level 4 trains in operation, from airport transportation shuttles to entire metro lines. In the world of train operation, fully autonomous vehicles are already here and in widespread use.
However, while fully automated, many of these systems aren't truly using AI or machine learning. That's changing though as the future of AI in transportation will help keep an eye on the tracks, crew and passengers. Computer vision systems are being used to conduct regular inspections of rails and train systems, providing visual inspection of anything that might be cause for concern. Computer vision is also being applied to keep watch on pedestrians or passengers who might encroach on tracks or pose safety or security issues. GE Transportation is using front-facing cameras to assist with train safety as well as track inspection. Similarly, machine learning is being used to provide predictive maintenance and advance warning of potential issues that may arise on the tracks.
AI is also helping train operators optimize operations. The AI systems help manage the combination of high-speed passenger trains and slower freight trains, reducing the inevitable delays that occur when a train falls behind schedule. The practice of Condition Based Maintenance uses big data to help optimize and predict when maintenance needs to happen on trains. Thanks to the rapidly expanding scale of manufacturing and asset maintenance industries, companies are now adapting to the wider applications of advanced algorithms on consumer-generated big data.
AI making airplanes more intelligent and automated
Airplanes have long been on the bleeding edge of adopting AI in transportation, outfitting planes with automation and intelligent technology. Autopilot technologies, fly-by-wire controls, and automated flight systems were first implemented in cockpits decades ago, and planes are practically flying themselves at this point. However, as recent events show, much can be done to make these systems more intelligent and more responsive to changes.
AI and machine learning are being applied in the cockpit to help reduce workload, handle pilot stress and fatigue, and improve on-time performance. There's no shortage of data available in aviation, from sensors, radar, control tower and satellite data. Machine learning systems are able to leverage Automatic Dependent Surveillance Broadcast for traffic situational awareness, the Maneuvering Characteristics Augmentation System (MCAS) to improve aircraft behavior and response to changing conditions, and runway overrun protection system to help prevent accidents at the airport. The Airbus A350 XWB aircraft has over 50,000 sensors collecting flight and performance data, generating over 2.5 TB of data each day.
Of course, what gets people excited in the field of aviation is the use of drones and fully unmanned autonomous vehicles (UAV) for all sorts of applications. UAVs and drones are being adopted to inspect fields and buildings, for shipping and delivery applications, on the battlefield, and many additional use cases. In these situations, AI and machine learning play a central role in the ability to control the craft, provide the computer vision for the application task, and interact with the real world. Aircraft companies are looking to scale up this sort of UAV technology to craft that can shuttle people around as well. Time will tell if the safety of such systems will reach the point where people will trust the lack of a human pilot in the cockpit.
On the ground, AI and machine learning are being used to help with flight control and air traffic operations. Air traffic control (ATC) systems are increasingly leveraging natural language processing to interpret voice messages over noisy communication channels. The AI systems are providing guidance on weather and traffic data, handling some routine information requests, and otherwise helping to alleviate the workload and burden from traffic controllers. Airbus recently launched a public AI challenge called the AI Gym to help improve ATC operations.
In back office airline operations, AI and machine learning are greatly improving the ability of airline companies to deliver reliable, high quality results to their customers. At United Airlines, the company is investing in AI and machine learning to turn their enormous quantities of big data into better predictive analysis for maintenance, employee operations, customer demand and response to changing weather conditions. In fact, AI and machine learning have potentially more application in helping airlines deal with all this complexity on the ground than it does with the aircraft in the air. Airlines are increasingly making use of facial recognition technology to simplify and speed up the check-in process and handle increasingly more complicated security challenges.
The challenge to the future of AI in transportation
There are a number of challenges holding back the adoption of AI in transportation, especially in areas of public transportation. In particular, the use of AI-enabled MCAS systems has been implicated in the crashes of two advanced 737 Max 8 Boeing jets. While the fault might not lie specifically with the machine learning systems or algorithms, reduction in trust and faith from the public in AI-enabled systems will slow industry adoption of the technology. Commensurate with the adoption of AI technology needs to be increased use of transparent and explainable approaches to AI that will provide pilots and ground operations answers about why certain things are happening in real time, why certain decisions were made, and give humans the ultimate ability to control the craft in emergency situations.
Similarly, it's not just machines that need to be trained and learn from experience -- so too do humans. There's a great risk of over-dependence on automatic and autonomous systems. Human pilots and operators might not have enough real-world experience and intuition to know how to handle manual control of systems at critical, and stressful, moments when a human can provide the guidance needed to avoid potential catastrophe. In the same way that technology companies, transportation providers and operations organizations are investing heavily in AI and machine learning to improve their businesses in all the ways mentioned above, so too do we need to have investment to help humans learn how to interact with these systems better. The purpose of transportation has nothing to do with machines, after all. It's about helping us transport ourselves and our goods from one place to the other.