top of page

AI to decarbonize Mobility

What AI-enabled solutions have a chance to reduce CO2e emissions in Mobility significantly, in particular for the Greater Dublin Area?


Photo by Li-An Lim on Unsplash


'code red for humanity'

The recent Climate change: IPCC report is 'code red for humanity': fortunately, EU countries have arrived at a deal to set a more ambitious climate target of cutting emissions by at least 55% by 2030, compared to 1990 levels.


According to CO2 Emissions | Energy Statistics In Ireland, Ireland has a 20% Higher CO2-eq (CO2 equivalent: CO₂ and other Greenhouse gases) emissions intensity than European average in 2017.


Looking at the local Irish greenhouse gas emissions, Transport was responsible for a massive 20.3% of Ireland's greenhouse gas emissions in 2019. Not only that, between 1990 and 2019, the transport sector showed the greatest overall increase at 137.0%, with road transport increasing by 142.6%, according to the Environmental Protection Agency.




But how are we going to cut Transport emissions in such a short-time frame?


We are going to need massive investment, political and individual commitment and innovative, smart solutions: Artificial Intelligence will play a big part!





Challenge

Specifics of Irish Transport emissions

Looking closely at domestic Transport emissions in Ireland, over 95% is Road Transport, whereas domestic Railways, Navigation and Aviation transport are very little developed.

“Passenger cars were responsible for 57% of road transport emissions in 2019, with Light Goods Vehicles responsible for 15% and Buses and Heavy Goods Vehicles responsible for 27%.” Transport - CSO


Transport Emissions are projected to decrease by 13.4% over the period 2020 to 2030 to 9.1 Mt CO2-eq under the With Additional Measures scenario (shown as the line in the graph below), which assumes 936,000 electric vehicles will be on the road by 2030.



Ireland’s population has increased by 30% in the 1990-2019 period, a much higher proportion than the average EU population growth, both due to natural and migration population growth. A report from the Central Statistics Office has predicted that the population of Dublin could increase by 31.9% by 2036!


To achieve a 51 per cent emissions reduction by 2030 [as per the Climate Bill], significant new measures will need to be identified and implemented”, the EPA says.


Need to Prioritize


Quantifying the impact of proposed solutions is key to prioritize for such a challenging goal, however it is always difficult to model effectively. I have included recommendations here from the Shift Project - a French think tank dedicated to promoting decarbonizing the economy - report on Mobility for the bigger Paris area with existing GHG reductions evaluations.

In order to compare the proposed solutions for the Greater Paris area with the Greater Dublin area (40% of Ireland population), we should get an idea of the similarities and differences for each capital:


Population - total and density - source Census 2016 CSO and 2019 Eurostat


Paris intra-muros - with all its romantic film sets - is about the same size as Dublin city, but almost 5 times more dense! Paris is actually the most densely populated city in the European Union.


This density, as well as historical reasons, help to explain major differences in public transport:

  • Paris: Metro: 16 lines, 225,1 km network - Tramway: 10 lines, 115 km.

  • Dublin: Metro: 0 line. - Luas (Tramway): 2 lines, 42 km.


Solutions

The Irish Citizen Assembly settled on a list of recommendations for the State to tackle Climate Change, however we need to look as well as individual and commercial opportunities to reduce emissions in Transport. Here is a non-exhaustive list of recommendations where AI has a significant role.


The main proposed solutions here are around:

  • Transport Electrification

  • Shared “Last Mile” delivery

  • Télétravail

  • Micromobility: include bicycles, e-bikes, electric scooters, electric skateboards.

  • Express Public Transport

  • Carpooling

  • Mobility as a Service


Transport Electrification

Quantifying the impact of EVs on emissions

To quantify the impact of new electric vehicles (EV) instead of Internal Combustion Engines (ICE) on reducing GHG emissions fully, we need to take the full Life-cycle emissions, that is both the day-to-day emissions per km (petrol / diesel / electric) and the manufacturing emissions . We won’t go into other benefits like lower pollution here.

The good news is that when looking at an ICE vs. similarly sized Electric car, the total GHG emissions of the electric car is much smaller in Ireland. Electric vehicles are much more efficient emissions-wise provided the electricity is mostly produced with low carbon tech ( wind, solar, hydro, geothermal..), no coal plant generation ! In Ireland, as the percentage of wind electricity is going up, it’s definitely worthwhile.


For example, on How clean are electric cars? For a medium car : ​​


Source: transportenvironment.org


However in the first 20,000 km or so driven, the emissions per km is higher due to the large emissions for manufacturing the battery in particular.

Interestingly, the local Irish greenhouse gas emissions reported in the EPA figures above don’t include the manufacturing emissions, since those emissions are accounted for in the manufacturing country. They are estimated as “Consumption emissions” in other sources like Carbon Footprint Results but are difficult to validate.


For the purpose of doing a very rough estimate of GHG emissions savings for Ireland total emissions, taking a few assumptions:




Possible scenario to reach 1 M EV by 2030: aggressive target!


  • CO2 emissions per km (average over lifetime) based on medium car , e.g. VW Passat:

    • Electric emissions CO2 emission per km: 102

    • Diesel emissions CO2 emission per km: 233

  • Avoided Emissions for 1 million cars with an average annual mileage of 15,000 km per year : 1.965 Mt CO2-eq.

  • Emissions from road transport have remained relatively stable for the last 4 years, at an average 11.6 Mt CO2-eq.

The avoided GHG Emissions would be thus 17% of the current Road transport emissions, consistent with the EPA Projections above.


Note, hydrogen fuel cells for vehicles are also an option, however between now and 2030 the low level of production is likely to reduce it to hard to decarbonize industry and possibly heavy goods vehicles, shipping and aviation, as explained in In-depth Q&A: How will the UK's hydrogen strategy help achieve net-zero? International aviation and shipping is out of scope for this analysis, however hydrogen-fueled heavy goods vehicles may help in the short term.


EV Challenges and AI Solutions

The 1 M EV target is obviously very aggressive. Electric cars are still more expensive than Internal Combustion Engine cars, so the state provides a range of Electric Vehicle Grants.

Range anxiety is actually not as much a problem in Ireland as driving distances are relatively small and battery range is improving.


Charging anxiety

The number of charging points is still very low and needs significant investments.

  • Route Planner: with chargers by type on the way, and availability forecast.

  • Individuals can offer their home charge point for a fee to other registered users.

Will we have enough electricity?

With highly intermittent renewable electricity like wind power (an essential component of Ireland’s decarbonisation efforts) , the question is less “how much” but “when” will more electricity be needed. And the good news with EVs is that essentially they are batteries on wheels!


  • The installation of Smart Meters could reduce peak demand from domestic users by up to 8%: Electricity Smart Metering Customer Behaviour Trials (CBT) Findings Report , most importantly at peak times. In particular, Smart meters enable operators to control when a vehicle is charged, also known as V1G, to optimise use of renewable at low times and lower bills for the customers, with successful pilot schemes in the UK and France .

  • In fact, as an AI course project, we proved Predicting excess wind electricity in Ireland is possible with ML algorithms!,

  • Using fully charged EVs as a giant battery to give back to the grid when needed! In the future, Vehicle-to-grid (V2G), a system in which plug-in electric vehicles, such as battery electric vehicles (BEV), plug-in hybrids (PHEV) or hydrogen fuel cell electric vehicles (FCEV), will communicate with the power grid to sell demand response services by either returning electricity to the grid or by throttling their charging rate.V2G storage capabilities can enable EVs to store and discharge electricity generated from renewable energy sources such as solar and wind, with output that fluctuates depending on weather and time of day.

  • In your home, zappi is a smart EV charger from myenergi with advanced optimisations. Not only does it operate as a standard EV charger, but it also has optional charging modes to prioritise energy generated from your Solar PV or wind generation.

But we need more reductions!

The suggestions below only address personal mobility as opposed to commercial mobility.


Shared “Last Mile” delivery

The proposed solution here is to completely REPLACE consumers driving to a supermarket with consumers ordering their weekly shopping online and shopping delivered effectively through shared rounds for closely located customers.

This solution makes a number of assumptions:

  • Orders deliveries are grouped and delivery times are pre-agreed with a time range.

  • Local multi-services multi-brand points are available to pick up parcels when consumers are not at home and bring back recyclable packaging, supporting people who have difficulties ordering online…

  • Delivery vehicles must be low carbon.


Left graph: 500 individual journeys to a large supermarket (in the center of the figure), generating approximately 13,200 km of movement. Right graph: Shared deliveries replacing the 500 individual journeys to the supermarket, generating around 2,080 km of travel. Source: theshiftproject.org


The Reduction in CO2-eq emissions Life Cycle is estimated for the Ile-de-France at 5%, which should match the conditions in the Greater Dublin Area.


A number of AI Solutions already enables such grouped deliveries, for example the GraphHopper Directions API with Route Optimization

“If you know the travelling salesman problem, then you already know a small subset of problems that can be solved with our route optimization software. If you need to route a fleet of vehicles (or workers) to deliver items or services to your customers, try integrating our route optimization into your application. It assigns routes to vehicles so that total transportation costs are minimized, and it can consider an arbitrary number of business-specific side constraints like time windows, driver skills, vehicle capacities and more.” And Reduction in CO2 emissions Life Cycle !

  • Optimal routing for multiple deliveries / repairs.

  • “Eco-driving”, reducing Idling time.

  • Diagnostic: tyre pressure sensors.


The Promise of Low-Carbon Freight with cargo bikes in London is a great case study of an efficient implementation!


Remote Working

As the current Covid-19 crisis has proved, remote working is possible for a lot of domains. Most CEOs anticipate hybrid working model - KPMG

In the Shift “Max” scenario, “we have assumed that 47% of employees telecommute 2 days per week. Trips on remote days are eliminated: the employee stays at home (for 50% of telework days) or goes to work in a third place accessible on foot or by bike (for the other 50 percent)”.


The “max” scenario is likely to apply here - with a high proportion of tech. workers and a cultural shift since Covid - which would mean a reduction in CO2 emissions of 8%.


Remote working AI-enabled tools: video conference with real-time audio transcription, messaging apps with translations, … This section could easily be a series of articles in itself and I will point you to other articles, even far fetched ones, Facebook Launches New App That Turns Virtual Meetings Into A Dystopian Nightmare !



The Cycling System and other Micro-mobility solutions



The Shift Project looked at a holistic “Cycling System” rather than just Cycling Lanes, to mirror the current “Car System” which provides not only road infrastructure, but also security and prioritisation through traffic lights, bridges, etc… As well as extensive Parking and repair networks, and strategic access to public transport hubs and sharine capabilities.


Depending on the speed and extent of delivery of the “Cycling System” , the Shift reports estimated a reduction in CO2 emissions between 6 and 24% for the Ile-de-France. However, a short cycle to a train station with fast and reliable connections is a key use case and as we have seen above, this is rarely available in the Greater Dublin Area and requires long term investment, so we’ll assume a middle of the road by 2030: 15%.

Interestingly, the micro-mobility (scooters, bikes rental) providers integrate such “holistic system” needs from the get-go, thus leading the way in AI in order to optimise scooters / eBikes pre-positioning, data sharing for local multimodality solutions and efficient recharging amongst others.


There is already a large number of providers waiting on the passing of the Road Traffic (Miscellaneous Provisions) Bill, which will legislate for the use of e-scooters on Irish roads and pave the way for e-scooter sharing companies to launch operations, including Irish e-scooter and bike sharing start-ups such as Zipp Mobility and Bleeper and international players such as Zeus, Voi, Lime and Dott according to https://www.siliconrepublic.com/start-ups/bird-e-bike-e-scooter-plans-ireland .

Also, a computer vision tech trial in DCU aims to increase safety for scooters and other road users: equipped with the Luna technology, the Zipp scooters are capable of running pedestrian detection and lane segmentation algorithms, allowing the vehicles to understand how many people are in their path, as well as whether they are on the road, a cycle lane, or footpath.


Express Public Transport

The “Express Public Transport” consists of the express bus and the commuter train lines, particularly suited to “commuter” mobility. Express buses typically have few stops and take high-speed roads, so that their speed is competitive with a private car. The Madrid Interchange Plan is a role model for implementing such complementary modes efficiently with Mobility simulators to validate the design and real time integrated data to provide users with clear, intuitive and efficient information.


While the commuter trains are already significantly used in the Paris region urban areas, in the Dublin area commuter trains are limited and express bus lines are extremely limited .

BusConnects Dublin includes the Network Redesign and the Core Bus Corridors plans for the greater Dublin and is an essential part of the Climate Action Plan 2019, however it is not planning Express buses. MetroLink and Dart+ are under public consultation and won’t be completed before 2027.


The Shift Project report estimated a reduction in 7% CO2 emissions for improvements in Express Public Transport the Ile-de-France, however in the greater Dublin area, given the low level of commuter trains and uncertainties about planning, it is really difficult to estimate possible reductions.


In and around Dublin, improvements in bus lines frequency, speed, reliability and interconnections would be the Quickest Win, as infrastructure requirements are minimum.

Examples of AI companies supporting public transport:

  • An Irish company CitySwift: The bus data engine provides analytics and route optimisation for bus companies through predictions depending on traffic patterns, demographic, weather, special events…

  • Transport prioritisation including support for population at risk : e.g. enforce a new bus lane with Pros and Cons for a local council or create a new route like Local Link Cork’s route 254 with Demand Responsive services : Remix Delivers Better Transport Service to Rural Areas through a collaborative mapping platform for transportation planning and decision-making.


Carpooling

Carpooling is the sharing of car journeys so that more than one person travels in a car, and prevents the need for others to have to drive to a location themselves.

It can be as old school as a parent driving their kids to school to BlaBlaCar with more than 90 million members that travel by carpool or long-distance buses, to apps like KAROS – a dynamic and predictive short distance carpooling route planner services, enabling commuters in France's Paris region to get to work faster, save on costs and meet new people. And reduce carbon emissions!

The Covid pandemic has obviously put huge breaks to carpooling efforts. In the longer term,

there can be a social reluctance to share a car with a stranger anyway. Transport for Ireland has sensible recommendations for businesses looking to set up a Carpooling scheme , for example with Kinto carpooling app.

On the other hand, Transportation Network Companies (TNCs) like Uber, Lyft, and Via started offering ride-sharing to share costs for a taxi ride: UberPool and Lyft Shared. Free Now Match launched in 2019 in Europe.

Companies like Ecolane provide local authorities with next-generation scheduling software to optimise shared rides to doctors’ appointments and the like.

The big advantage of these offers is that they can immediately use the existing car-focused infrastructure. However, significant incentives are required for both drivers and passengers for such offers to expand significantly, like High-occupancy vehicle lanes.

For the Ile-de-France, the Shift reports estimated a LifeCycle CO2 emissions reduction ranging widely from 2% to 16% where 16% assumes ALL possible carpooling opportunities are used, for a rough estimate for the Greater Dublin Area we’ll take the middle value: 9%.


Mobility As A Service MAAS


Putting it all together! Photo by Alvaro Reyes on Unsplash


Multi-modal optimization, a.k.a. Mobility As A Service (MaaS), provides optimal options for users trying to get from A to B, with a number of constraints: time, cost, walking speed, user preferences, accessibility requirements.

The MaaS platform - either through a white-label local authority app or their own brand - will offer options combining all transport modes, from public transport to bike-sharing and taxis, possibly offering a once of payment per trip / subscription.

Ultimately, Autonomous cars should be part of the mix !


MaaS providers with an mobile app, as well as the Web interface, provide users with up-to-date options, as they have access to a wealth of data:

  • The user’s trip intention: where the user wants to go from where they are, and when.

  • Which option the user actually takes, inferring walking for example from the location updates and motion sensors (accelerator..)

  • Updating other users on bus delays accurately from both data sources: transport operators location updates and users with the app on that bus.


Pros

  • Improve Accessibility

  • Done right, Local Authorities can manage efficient mobility for cities at a reasonable cost and low carbon impact.

Risks

  • Proprietary data - providers lock-in

  • Opaque algorithm with conflicting requirements: transport operators profitability, low carbon optimisation...


Many MaaS providers are emerging, for example:

  • Whim app of MaaS Global in Helsinki launched back in 2016 made more than 16 million trips.

  • Moovit - recently acquired by Intel - is used by 800 million users in 3,100 cities in more than 100 countries. The app allows multimodal trip planning that combines public transit, bicycle and scooter services, ride-hailing and car-sharing.

  • SkedGo offers MaaS services through their API, SDK and White Labels with many unique features, such as accessibility routing, occupancy information and personalisation options (tailored route prioritizing low carbon, exercise, speed of trip, cost of trip or convenience).

  • BMW-Daimler REACH NOW MaaS app deployed in 48 German cities show car manufacturers realization they need to evolve the Business Model from car ownership

  • Uber Transit Multimodal options make it easy to plan a trip end to end with UberX and public transit. Uber AI leverages ML models powered by neural networks to forecast rider demand, pick-up and drop-off ETAs, and hardware capacity planning requirements, among other variables that drive their operations.

  • A pilot project in Dublin consisting of 4 organizations - Enterprise Rent-A-Car, UCD, Bleeperbike and Good Travel Software (GTS) – went into operation last May as a smart mobility hub at Dublin City Council’s Wood Quay office. GTS provide a bundle of AI-based services to coordinate the hub for:

    • Car sharing: Demand Prediction, dynamic pricing to ensure vehicles are where they need to be.

    • Dynamic Shuttle: Dynamic adjustment of shuttle trip to accommodate addition of passengers...

  • Let’s not forget Google Maps!

It is too early to estimate the reduction in LifeCycle CO2-eq emissions, however with the right governance, MaaS platform should enable the highest range for most of the solutions above.



summary

  • A prerequisite for substantial CO2-eq emissions reductions is a need for a substantial and fast public infrastructure investment, as well as a cultural shift towards “lighter” mobility.

  • All reductions above don’t add up, for example, reductions thanks to Electrification and Carpooling will overlap.


Rough Estimates of Max possible CO2 Emissions reductions in Transport


Call to action

  • As an individual:

    • Do you need to take the car or can you cycle or take public transport?

    • Can you recommend alternative solutions to your friends / network?

    • Can you organize carpooling with your colleagues to go to work? - Keep safe, with masks :=)

  • As a voter:

    • What parties have solid policies to invest massively in low carbon infrastructure? Vote for them!

    • Join https://www.dublincommuters.ie/ to lobby with like-minded residents .

  • As an entrepreneur:

    • What opportunities are open to help transport users / transport operators / local authorities ? Go for them!

    • Browse through the open datasets in Dublin's Open Data Portal , find new ways to use data to predict and optimize transport!

Please, add comment to this blog to add projects where AI is making a substantial dent in Carbon emissions in Transport!


Author: Catherine Lalanne https://cath709.medium.com/


0 comments

Related Posts

See All

Kommentare


bottom of page