Search Results

Centre of Excellence logo Return to all Events TfSE CoE & Arup | Impacts of Devolution on Buses Transport for the South East (TfSE), in partnership with Arup, invites you to a webinar exploring the future of bus governance, planning, and infrastructure in light of the Government’s English Devolution and Community Empowerment Bill. Who should attend: Local Transport Authority officers in the South East with responsibility for bus planning and operations. What will it cover? This session will provide insight and discussion on the potential implications of the Bill for the management and funding of bus services and infrastructure across the South East. Drawing on recent TfSE and Arup research, the webinar will focus on three priority areas identified by local authority officers: •Devolution and Bus Governance – What devolution could mean for the way Local Transport Authorities (LTA) manage buses, infrastructure, and cross-subsidy decisions. •Cross-Subsidy Planning – How LTAs can approach financial planning for services in the context of devolution. •Cross-Boundary Infrastructure – Addressing the challenges of funding and managing bus infrastructure that spans multiple authority boundaries. The webinar will include opportunities for questions and discussion with colleagues across the South East. Associated Documentation 2025-09-30

A tool to calculate carbon emissions for operational, construction and maintenance activities undertaken on behalf of National Highways. Back to Key Tools National Highways | Carbon emissions calculation tool A tool to calculate carbon emissions for operational, construction and maintenance activities undertaken on behalf of National Highways. Explore tool

The Walking and Cycling Index supports leaders of cities and towns to understand and improve walking, wheeling and cycling across the UK and Ireland. It provides high-quality evidence to help bring our neighbourhoods back to life and ensure walking and cycling are attractive and accessible for everyone. Back to Key Tools Walk Wheel Cycle Trust | The Walking and Cycling Index The Walking and Cycling Index supports leaders of cities and towns to understand and improve walking, wheeling and cycling across the UK and Ireland. It provides high-quality evidence to help bring our neighbourhoods back to life and ensure walking and cycling are attractive and accessible for everyone. Explore tool

The EVCI Framework takes a whole systems approach in identifying the significant requirements placed on the electricity grid and energy networks arising from the electrification of road vehicles. To view TfSE data, please navigate through the top panel. Back to Key Tools STB EV Charging Infrastructure Framework The EVCI Framework takes a whole systems approach in identifying the significant requirements placed on the electricity grid and energy networks arising from the electrification of road vehicles. To view TfSE data, please navigate through the top panel. Explore tool View the Statement of Methodology here

Return to all Resources TfSE | Horizon Scanning Briefing Note: Artificial Intelligence in Transport Planning A bite-sized, actionable overview of Artificial Intelligence (AI) in Transport, with a specific focus on transport planning in the South East. Three key takeaways Artificial Intelligence (AI) is already delivering in transport: from arrival-time prediction, scheduling and adaptive signal control to predictive maintenance and consultation analysis. These tools speed up insight and can support decision making, while keeping human judgement in charge. The toughest barriers to deployment aren’t technical – they’re organisational and data-related. Fragmented and low-quality data, restricted sharing, legacy procurement, skills gaps (especially data science and assurance), and integration with existing systems slows progress and erodes trust without clear governance and monitoring. This briefing note recommends three immediate actions for those interested in deploying AI in their organisation: Pick a high-value use case and get the basics on data right before you start; Pilot → evaluate → deploy with proportionate assurance and transparent reporting; and Build mixed teams to deploy AI, and align with the DfT Transport AI Action Plan to access guidance, datasets and opportunities to develop. Introduction This Horizon Scanning Briefing Note intends to give a bite-sized, actionable overview of Artificial Intelligence (AI) in Transport, with a specific focus on transport planning. As such, it represents a snapshot in time of the development and application of this technology. It also intends to give some actionable key insights that are of use to transport planning professionals across the South East. AI is a vast and fast-developing field of technology. Therefore, this briefing note is intended as a starting point for the reader to explore the area further. It includes links to other research reports, web pages and documents, which the reader is encouraged to read themselves to further their understanding. What is Artificial Intelligence? AI has many definitions and can be understood in a wide variety of ways. In its broadest sense, artificial intelligence means computers (software or hardware) performing tasks associated with human intelligence. This includes recognising patterns, learning from data, making predictions, and choosing actions. John McCarthy , computer scientist at MIT and Stanford and widely referred to as the ‘father of AI’ called AI “the science and engineering of making intelligent machines.” Russell and Norvig took this one stage further to highlight the reasoning aspects of AI, saying AI is “the study of agents that receive percepts from the environment and perform actions.” The Chartered Institution of Highways and Transportation (CIHT) defines AI as the ability of technology to “perform human-like tasks, such as perception, logic, and reasoning.” It is distinct from a similar technology, machine learning, in that “machine learning allows an AI system to autonomously learn about and develop the task it has been given, based on self-learning algorithms, statistical models, and data it has been fed.” Consequently, whether a system is defined as AI is open to interpretation. A commonly referred to ‘test’ of whether a system is AI is the Turing Test . Proposed by Alan Turing in 1949 as the ‘imitation game,’ this is a test of whether a machine can exhibit human-like intelligence by having a human be unable to distinguish whether, in a text-based conversation, the response was written by human or machine. While ‘passing’ such a test does not formally classify a system as AI, it provides an indicator as to the intelligence of a computer system. Common elements of AI systems Whilst there are a variety of definitions of what AI is, there is some degree of commonality when it comes to elements of systems that are generally considered to be either AI or AI supported. These include the following: Data-driven – AI systems rely on large and diverse datasets (structured, semi-structured, or unstructured). The quality and representativeness of that data directly shape outcomes (hence the common phrase “rubbish in, rubbish out”). Inference and learning – Instead of being programmed with fixed rules alone, AI systems infer patterns and relationships from input data to produce outputs such as predictions, recommendations, or decisions. Many use machine learning to improve performance over time, though as noted by the CIHT machine learning itself is not AI. Goal-directed outputs – They are built to achieve explicit or implicit objectives: for example, predicting demand, detecting anomalies, or optimising routes. Autonomy and adaptiveness (to varying degrees) – Systems can operate with minimal human intervention, and some adapt their behaviour after deployment. Others can do this, but only when permitted to do so by the human in control. Close resemblance to human cognitive functions – They perform human-like tasks such as perception (recognising images or sounds), reasoning (logic, planning), prediction (forecasting events), and generation (creating content or decisions). Environment interaction – Many AI systems sense, analyse, and act in physical or digital environments (as indicated by Russell & Norvig’s definition outlined previously). While some require prompts, others have some degree of autonomy to do this. Explainability and accountability needs – Because outputs influence the real world, AI systems require safeguards like documentation, monitoring, and mechanisms for human oversight. The degree to which these are deployed varies between systems. For example, a customer may not have full oversight of an AI deployed by a supplier. Current uses across transport and transport planning There are numerous examples of how AI is being deployed across transport and logistics. Many such systems are commercial, and as such the impacts of the deployment of these systems is not publicly available. Some systems either have some or all of their source coding published openly, depending on the organisations responsible for developing them. Applications in transport planning AI is being applied to several areas of transport planning. A notable example is scenario modelling and digital twins . Planners are fusing sensor, ticketing and land‑use data to test operational and spatial scenarios—for example, the reliability effect of small timetable changes or the network implications of new changes to road and walking and cycling networks. Natural‑language processing (NLP) is helping analyse consultations by turning thousands of free‑text responses into auditable, traceable findings in hours. UK authorities use tools like FutureFox’s ConsultAI to cluster themes, quantify sentiment and pull verbatims for statutory reporting—accelerating iteration while keeping human judgement in the loop. Another area is equity analysis . Privacy‑preserving datasets can augment sparse samples (e.g., on low‑usage corridors) and support equity testing without exposing personal data. Planners are beginning to use these approaches to compare distributional impacts across groups before consultation. A final emerging area is using Generative AI (which is AI that produces an output) in business‑case drafting and technical reports. Platforms such as SchemeFlow support engineers and planners by drafting transport assessments and business‑case sections from local policy, datasets and engineer prompts. Content is structured to match common templates and is reviewed/edited by professionals, reducing turnaround from weeks to days on scoping and early drafts. Other transport applications From a scan of the literature and the web, there are several other key use cases in transport have been defined. This list is by no means exhaustive, as new examples of AI deployment are being added all the time. Public Transport Arrival-time prediction & real-time information . Operators use machine-learning models to predict bus and train arrival times from data feeds, historic running and disruption data. Demand forecasting & revenue/yield . AI is used to forecast demand by stop/route and to support dynamic pricing or targeted promotions (where appropriate). This builds previous extensive use of machine learning on smartcard, ticketing and survey data to understand traveler behaviour and optimise service plans. Timetable, vehicle and crew optimisation . UK bus operator examples show AI-assisted scheduling lets planners generate and compare multiple data sets in minutes. This can further improve timetables to optimise the use of vehicles and drivers. Passenger information & personalisation . Recommendation models and conversational tools can tailor journey options and disruption messages based on user context (e.g., step-free routes, crowding). Industry analyses highlight growing use of decision-support to improve customer experience in such ways. Fraud detection & anomaly spotting . Supervised and unsupervised models flag suspicious usage in smartcard data (e.g., unusual tap sequences or device sharing), complementing traditional revenue protection. Paypal, for instance, uses AI to detect potential fraudulent transactions in real time. Predictive maintenance . Computer vision and sensor analytics spot rail and fleet issues earlier, reducing downtime and improving reliability . Traffic Management and Control Adaptive signal control (corridor/junction). AI systems learn demand patterns and adjust stages and splits in real time. In Cambridge, VivaCity’s Smart Junctions pilot used AI and multimodal computer-vision sensors to adapt signals, detect pedestrians/cyclists and prioritise sustainable modes while maintaining motor traffic performance. Incident detection & response . Vision models that analyse CCTV and probe data can reduce detection time and improve actionable alerts for control rooms. Safety analytics & near-miss detection . UK pilots are using AI safety analytics from roadside sensors to prioritise remedial measures on the strategic and local road networks. Logistics Warehouse automation & robotic picking . Ocado uses AI-enabled robotic systems to automate picking and coordination in fulfilment centres, boosting throughput and consistency. AI-assisted routing, inventory and labour planning . Retail and parcel operators apply demand sensing, route optimisation and shift planning to cut empty running and dwell times —improving service reliability and emissions. Decarbonisation . AI supports electric charge-point siting, vehicle-to-grid scheduling and ‘green’ routing; UK professional sources highlight EV infrastructure planning and smart-logistics use cases as current areas of value. Learnings from this early experience AI is already helping planning teams – but it isn’t replacing professional judgement. Used well, these tools free up time for the human work of prioritising, explaining trade-offs, and making fair, defensible decisions. Some key learnings include: Faster insight. Instead of waiting weeks for a bespoke study, teams can now combine ticketing data, live vehicle locations and mobile sensing to produce demand and reliability views in hours. For example, this means you can spot bus-bunching hot spots, or test whether a small timetable tweak would improve operations on a corridor. Better engagement evidence . Consultations often produce thousands of free-text comments. AI helps turn this into structured evidence, identifying common themes, highlighting emerging issues, and pulling representative quotes. This can make consultations more accessible to decision-makers and communities alike. New approaches to appraisal . AI can support more systematic testing of resilience and equity. You can simulate disruptions (e.g., a temporary road closure) to understand knock-on delays and where targeted mitigations would help most. You can also explore distributional impacts – who gains and who loses – before schemes are finalised in a more dynamic fashion compared to models developed many years in the past. Skills and operating model. Early adopters report the biggest gains when mixed teams work together: transport planners and engineers for context, data scientists to deal with data and the systems, product or project managers to keep users and outcomes in focus, and assurance leads to check fairness, privacy and performance. A lightweight AI assurance process helps teams move quickly while staying safe and accountable. The Department for Transport’s Transport Artificial Intelligence Action Plan The Department for Transport’s Transport AI Action Plan (DfT) sets out how the department and the wider sector will adopt AI responsibly and at pace. The plan’s objectives are delivered through five work-areas: Leadership (policy, guidance, governance), Skills & Capabilities (upskilling DfT and the sector), Infrastructure & Data (access, stewardship, compute), Applications (prioritised pilots/use-cases), and Engagement (sharing practice across industry and authorities). Concrete actions include: Creating a Transport Data Action Plan (2025) to improve discoverability and reuse of priority datasets for AI; Building guidance, procurement support and an “assurance” toolkit for AI projects; Establishing sandboxes and communities of practice; Backing pilots in areas such as adaptive traffic management, predictive maintenance and customer information Inside DfT, the plan commits to building internal capability (AI skills pathways, recruitment and training), better governance and oversight, and alignment with cross-government risk management approaches. Externally, it aims to help authorities and operators move from proof-of-concept to production with clearer data access, evaluation standards and post-deployment monitoring. The plan is explicitly tied to the UK’s “pro-innovation” AI regulation model led by the Department for Science, Innovation, and Technology. This asks sector regulators and departments to apply five principles (safety, transparency, fairness, accountability, contestability) without creating a single AI regulator. This is so transport guidance remains context-specific while interoperable with international regimes. The package addresses the main blockers—data, skills, procurement and assurance—and signals that DfT will provide the policy cover, tools and examples local bodies need to adopt AI safely where it adds public value. Barriers to adoption Experience across UK authorities and operators shows that technology is rarely the hardest part of AI adoption. The blockers are usually organisational, data and assurance related. Below are some of the most common barriers, but this is by no means a comprehensive list. Data fragmentation and quality . Operational data is often scattered across back‑office systems and contractors. Metadata is incomplete, quality is variable (e.g., missing GPS fixes, inconsistent stop IDs), and datasets lack clear owners. This slows re‑use, and analysis— especially for multi‑modal corridors . Data access and sharing . Even when data exists, teams may not be able to use it. Contracts can restrict onward sharing . Privacy teams may lack capacity to design proportionate mitigations. Finally, partners can be uncertain about lawful bases for linking datasets (e.g., journey data with demographics for equity tests). Legacy procurement and funding . Traditional procurement routes are optimised for buying large systems, not running small discovery/alpha pilots and iterating. Outcome‑based specifications and dynamic purchasing can be difficult to apply quickly. Annual budget cycles also make it hard to fund monitoring and model maintenance after it becomes operational. Skills gaps, especially in data engineer and assurance . Most transport teams have strong domain expertise but limited capacity for data science and engineering (cleaning, linking, standards) and AI assurance (bias testing, explainability, robustness and performance monitoring). Without these skills, pilot projects fail at hand‑off to operations or fail to meet governance standards. Bias and fairness . Models trained on historical data can inherit or amplify bias (for example, prioritising investment where data is most plentiful, not where need is greatest). Black‑box models are hard to scrutinise in public decisions. Authorities need to be able to explain the outcomes, bias testing and documentation so decisions remain auditable and contestable. Integration with legacy systems . Promising pilots can falter when models need to feed decisions into existing control systems (e.g. Urban Traffic Control). Interfaces may be proprietary or undocumented , and cyber, safety and resilience requirements add complexity. This makes end‑to‑end testing essential. Change management . People adopt what they trust . If planners, controllers or depot teams are not involved early, AI can feel imposed, especially if roles or schedules change. Co‑design, clear comms and targeted reskilling can build confidence, but are rarely deployed. Key actions to take AI is moving quickly, but transport planning remains a human‑led discipline. The most effective organisations start small, focus on public value, and build the capabilities and safeguards that allow them to scale. The three actions below are the most important things teams can do now to turn potential into measurable improvements for passengers, freight and places. Start with a clear use‑case and strong data basics Pick one operational problem with visible public value (for example, reduce excess wait time on a priority corridor, cut incident detection time on the ring road, or shorten timetable changes). Define a baseline and a target before you begin. Treat data as infrastructure: create a short inventory of the datasets you’ll need, name the owners, note quality issues, and agree access and privacy controls. Use open data standards and keep a simple data dictionary so others can reuse your work. Pilot → evaluate → deploy, with proportionate assurance Run pilots. Compare outcomes to your baseline situation. Log what data you used, the model or method chosen, and known limitations. Where people are affected, check for obvious bias (for example, whether reliability gains are concentrated in already‑well‑served areas). Keep humans in the loop for decisions with a public impact. If the pilot meets its target, plan the next stage: who will own the model, how it will be monitored, how it can be paused or rolled back, and how the benefits will be reported. Publish a short, plain‑English summary so partners and the public can see what changed and why. Build a mixed team and develop skills Pair transport planners/engineers with a data engineer or analyst, and nominate a product lead to keep users, outcomes and timelines in view. Create a short AI assurance checklist (one page) that project teams can use. Use procurement routes that support piloting and iteration, and include clear data‑sharing clauses. Invest in data skills for schedulers, controllers and analysts so AI tools augment their day‑to‑day work rather than feel imposed. Connect to national communities of practice and align with the Department for Transport’s Transport AI Action Plan so your work benefits from shared guidance, datasets and exemplars. Start where the evidence is strongest, measure what matters, and build only as fast as you can govern. That balance can help turn promising pilots into trusted services. Suggested further reading if you want to learn more If you want to learn more, the below reports are a useful starting point. Department for Transport – Transport Artificial Intelligence Action Plan (2025) : Sets out how DfT will work with industry to harness AI safely and responsibly, including actions on data access, skills, governance and targeted pilots. Chartered Institution of Highways and Transportation – The role of data and artificial intelligence in achieving transport decarbonisation (2024): Practitioner‑focused guidance on how to apply data and AI to reduce emissions across planning, operations and maintenance, with practical case studies from UK authorities and supply chain partners. International Transport Forum (OECD) – AI for Transport Authorities: Principles and Practical Guidance (2025): A report covering roles and mandates for public authorities, risk/benefit assessment, continuous monitoring, and organisational capabilities (including inventories of AI use). TRL – Bridging the gap: overcoming the barriers to AI adoption in transport (2025): Evidence‑based review of organisational and technical barriers (data sharing, interoperability, assurance) with actionable enablers and procurement considerations. Case Studies More and more case studies of the deployment of AI in transport are coming to light every day. Below are a few examples for illustration. SchemeFlow – business case and technical report drafting: an AI platform used by built‑environment and transport teams to automate pre‑construction and transport assessment reports, using templates, local policy and data, and engineer prompts. Turnaround reduces from weeks to days. Learn more: schemeflow.com Future Fox (ConsultAI) – consultation analytics: natural‑language processing turns thousands of consultation comments and surveys into structured, auditable findings in hours; draft reports that meet statutory requirements. Learn more: thefuturefox.com VivaCity Smart Junctions – traffic management: AI video sensors and reinforcement learning optimise signal timings at junctions and across corridors, improving bus reliability and reducing delays; compatible with existing UTC systems. Case study: Cambridgeshire Smart Junctions Ocado Intelligent Automation – logistics: AI‑enabled robotic picking and scheduling in UK fulfilment centres improves throughput and service consistency; illustrates predictive inventory and labour optimisation. Story: Ocado’s AI‑powered robotic arms West Coast Motors with Optibus – public transport scheduling: AI‑assisted crew scheduling reduces split shifts and speeds up timetable iteration, improving driver work‑life balance and service reliability. Case: Optibus case study Associated skills: Future Technologies, IT Capability, Transport Planning

Chartered Institution of Highways & Transportation Logo Return to all Events CIHT | Climates Webinar: findings & recommendations Over the past nine months, the CIHT CLIMATES project has been led by CIHT President, Professor Glenn Lyons. The initiative was designed to help professionals explore what may lie ahead and how we can meet the challenges of climate change. This webinar will provide a deep dive into the main findings and recommendations from the CIHT CLIMATES, followed by a Q&A with the project leader and current CIHT President Professor Glenn Lyons Associated Documentation https://www.ciht.org.uk/event/climates-its-findings-recommendations/

Transport for the South East Centre of Excellence logo Return to all Events TfSE Centre of Excellence | Business Case Development: How to ask the right questions to get an appropriate and proportionate modelling response The third episode in a series of webinars, covering a range of topics suggested by officers. The first 15 minutes will be a recorded presentation, and the last 15 will be an opportunity for Q&A with our Subject Matter Experts: Joshua Jiao and Josie Drath. Not able to make it? The recorded section will be posted on the Centre of Excellence. Associated Documentation

ADEPT logo Return to all Events ADEPT | Carbon Leadership Programme - Training 1 The ADEPT Carbon Leadership Programme is a programme supported by DfT that will fund local highway authorities to measure and reduce their emissions. The session will include essential information for local highways authorities who are planning to apply to be part of the next cohort. Associated Documentation

Department for Transport Logo Return to all Events Department for Transport | Transport Analysis Guidance (TAG) Conference 2025- Day 2 Day 2 of 3- Online only, There will be two streams of programming running in parallel. Both can be signed up for via the link. Programming Stream 1- 09.30 - 11.00: AMES Supporting Best Practice 11.00 - 11.30: Break 11.30 - 13.00: AMES Analytical Methods 13.00 - 13.30: Lunch 13.30 - 14.30: Hot Topic- Transport’s Impact on Innovation 14.30 - 15.00: Break 15.00 - 16.30 AMES Evaluation Programming Stream 2- 09.30 - 11.00: AMES Uncertainty 11.00 - 11.30: Break 11:30 - 13:00: AMES: Protecting the Environment 13:00 - 13:30: Lunch 13.30 - 14.30: Hot Topic- Opening the Black Box, DfT’s Transport Spatial CGE Model 14.30 - 15.00: Break 15.00 - 16.30: Sign up to Stream 1 to attend AMES Evaluation (optional) Associated Documentation

Chartered Institution of Highways & Transportation, Logo Return to all Events CIHT | Improving Health through Active Travel workshop CIHT are hosting a workshop to gather insights from professionals in the sector. The discussion will help shape our response to key questions, including: 1.What prevents the successful implementation of active travel schemes? 2.How can we overcome these issues? 3.What data do we have that shows which initiatives work best? e-mail through the link below to sign up Associated Documentation mailto:technical@ciht.org.uk

Transport for the South East Centre of Excellence logo Return to all Events TfSE Centre of Excellence | Business Case Development : How to review a business case (and the key questions to ask) The second episode in a series of webinars, covering a range of topics suggested by officers. The first 15 minutes will be a recorded presentation, and the last 15 will be an opportunity for Q&A with our Subject Matter Experts: Mat Jasper, Steve Hunter. Not able to make it? The recorded section will be posted on the Centre of Excellence. Associated Documentation

Return to all Resources Department for Transport | Local highways maintenance appraisal tool The highways maintenance appraisal tool helps local highway authorities assess the economic costs and benefits of proposed maintenance. https://www.gov.uk/government/publications/local-highways-maintenance-economic-costs-and-benefits-tool Associated skills: Developing business cases