Not so Scary After All

For members of those generations that can still recall capably managing their personal and professional lives before the dawn of the digital age, artificial intelligence (AI) was introduced to us, courtesy of Hollywood and a trademark Austrian accent, as something terrifying and apocalyptic. Exactly 40 years later, in this era of remarkable innovation and proliferating investigation, the greatest risk to industry veterans and the mathematically indifferent (myself included), is that of glazing over at talk of learning vector quantization and Naïve Bayes. Undeniably, however, change is the perennial key to progress, and AI clearly holds the potential, in a public transportation context, for so much more than has been achieved to date.

Current Applications and Barriers to Adoption

Acknowledging that AI is still very much in its infancy, it comes as no surprise that public transportation applications to date are thin on the ground, and largely limited to pilots and trials. Given the unacceptable risks of rushing AI technologies to market too quickly, as highlighted by the tragic death of Elaine Herzberg in a driverless vehicle accident in 2018, critical lessons and learnings have still to be captured to ensure public safety, and ultimately, to gain public trust. The ideas, however, are in an abundance, spanning areas including network optimisation, asset maintenance, operational efficiency and safety.

In Singapore, the Land Transport Authority is reaping the benefits of AI in traffic management, through its integrated ‘Intelligent Transport Systems’ solution. Comprising a network of data collection technologies, including sensors and cameras, traffic data is analysed real-time by AI to optimise signal timings and manage congestion. For what is the third most densely populated country in the world, the resulting road network efficiency translates to reduced wait and overall travel times for commuters, plus safer roads.

On the London Underground, British Transport Police is collaborating with Transport for London to trial AI technology for revenue protection and passenger safety purposes. Utilising AI in conjunction with CCTV cameras to analyse video footage in real-time, suspicious behaviour and potential fare evaders can be alerted to station staff. Subject to successfully navigating expected privacy concerns, the solution could be deployed across the entire network, enabling increased security for patrons and reduced revenue loss.

In Germany, Deutsche Bahn is making a significant investment in AI and robotics technologies to optimise the maintenance of its InterCity (ICE) train fleet. Utilising new ‘E-Check’ technology, AI is tasked with determining deviations from the target condition of a train based on inputs from a camera and underfloor device. In only five minutes, the solution can identify the smallest inconsistency or need for a major repair on a 374-metre long XXL ICE, effectively revolutionising maintenance by minimising costs, reducing downtime and increasing passenger safety.

In Australia, Transport for New South Wales, in partnership with Cisco, recently trialled the symbiotic use of AI, the Internet of Things and Edge computing to obtain real-time views of bus, ferry and light rail services in Sydney and Newcastle. Providing valuable insights into customer comfort and demand, in addition to vehicle movements, adoption of this technology would support transit agencies to make informed future network decisions, towards improving the reliability of public transport.

The reality is that all public transit agencies are positioned to benefit from AI. The obvious barrier is investment. For those agencies already collecting patronage and system performance data via smartcard or other automated fare collection (AFC) solutions, the opportunity to realise increased operational efficiencies is within reach. In conjunction with investment in infrastructure, the applications of AI are then checked only by the pace and progress of scientific and technological advancements, not to mention regulations and legislation. In respect of an optimised customer experience, however, one path is clearly the adoption of account-based ticketing (ABT), which akin to AI, is also in its early stages. Could the basis for truly personalised and frictionless travel through ABT and AI be closer than we think?

COVID as Catalyst for Innovation – A Transition from ‘Contactless’ to ‘Touchless’?

Obliged to implement measures to prevent the spread of COVID-19, transit agencies turned not only to revised policies and practices, but also the adoption of new technology. Whilst the pandemic may have receded, travel patterns have changed indelibly. In consequence of remote and hybrid working, contactless payments, where a choice for patrons, are understandably favourable to smartcard use. Innovation, however, has provided for the next level of convenience, in ‘Touchless’ payments. Swiss based FAIRTIQ has introduced a check-in/check-out app which completely removes the need for interaction with AFC infrastructure or personnel. Already adopted across six countries in Western Europe, and drawing on GPS and AI technology, the solution may quickly be eclipsed by be-in/be-out implementations, removing the need even to use an app while on the move. Having addressed the call for seamless and safer travel during the pandemic, this model of account-based ticketing presents transit agencies with immense savings opportunities, negating the need for validation and vending infrastructure. True, there remain technical challenges in respect of in-transit communications and connectivity, but these will likely be easier to overcome than privacy hurdles associated with biometric solutions, such as that recently trialled on the Ui-Sinseol light rapid transit line in Seoul, South Korea.

Realising the Full Potential of AI

Clearly, we are only scratching the surface in terms of realising the benefits of AI in public transportation. The COVID-19 pandemic compelled the public and private sectors to look deeper into what is possible, leading to innovative concepts and solutions, most notably ‘Touchless’ payments.

In conjunction with the implementation of account-based ticketing, public sector investment in AI technologies and requisite supporting infrastructure are undoubtedly the vehicle for an optimised customer experience. Paradoxically, in this era of reduced and less predictable patronage, and compounded by intermittent pressures to reduce public sector debt, the promised return on investment of AI via cost savings may even be necessary to maintain the very sustainability of public transportation.

The cherry on the cake for the customer experience, however, requires a turnabout in prevailing public sector thinking, one which opens the way for the perceived holy grail of mobility to be achieved. Were public transit agencies to focus their energies and funding on the synthesis of account-based ticketing and AI, and make the fruits available to be consumed by others (subject, of course, to accreditation and regulation), will it finally become feasible that Mobility as a Service (MaaS) could be organically born of the free market? Offering patrons the opportunity to opt their transit agency accounts into MaaS schemes means benefitting from promos and personalised offerings (including dynamic ticketing products), and ultimately, choice of service providers, which fundamentally underpins all contemporary notions and narratives of mobility. A reciprocal sharing of data under such schemes would enable the continuous realisation of efficiencies and optimisations, providing not only for a harmonious marriage of the private and public sectors, but also a ‘truly optimised’ customer experience.

However, we know how difficult this has been to achieve to date, with multiple players competing for the same account and with differing business objectives. Maybe, AI can help us provide a path towards interoperability and an optimised customer experience, and assist in navigating around the obstacles that have hindered us so far?

Building on these concepts, questions to be explored in follow-up articles include:

• · How can interoperability be realised through AI?
• · In what ways can AI be utilised to optimise ABT implementations?
• · By what means can the customer experience be further enhanced by AI?