1.
Autonomous Shuttles-as-a-Service
(ASaaS)
Challenges, Opportunities, and Social Implications
Antonio Bucchiarone, PhD
MoDiS, FBK
bucchiarone@fbk.eu
https://bucchiarone.bitbucket.io/
24 November 2020

2.
Introduction
page
02
§ Mobility of people and goods in the center of transportation planning
and decision-making of the cities of the future.
§ Electric Vehicles and shared mobility to accelerate the transition to
zero-emission vehicles and maximize climate and air quality benefits.
§ Cities are introducing mechanisms to encourage citizens to change
their behaviours and to make sustainable mobility habits daily.
§ Autonomous Vehicles are capable of moving without the full control of
humans and focusing on sensing their environment and automate some
aspects of safety (steering, braking) without human input.
https://urban.jrc.ec.europa.eu/thefutureofcities/mobility/#the-chapter

3.
Introduction
page
03
§ To reduce carbon production and exploit the time spent inside the car.
§ Drivers can be able to employ their time to do other activities instead of
only driving.
§ To reduce traffic, congestion and accidents that are caused by driver
errors, fatigue, alcohol, or drugs.
§ Last-Mile Mobility: movement of people and goods from a transportation
hub to a destination.
§ Home delivery efficiency in the Logistic Domain: spatial dispersion of
the parcels’ recipients and the frequency of failed deliveries.

4.
Autonomous Shuttles
page
04
§ From this perspective, the growth of autonomous shuttles in urban
public environments could enable new services to deal with the new
challenges posed by large cities, which require the combination of the
mobility of people and goods.
§ Several pilot experimentations prove significant technology development
results, as well as the citizens’ acceptance in many cities all over the
world, in countries such as:
§ Germany, France, Switzerland, Finland, Sweden, The Netherlands,
and Estonia.
§ We introduce the concept of Autonomous Shuttles-as-a-service
(ASaaS) as a part of the Mobility-as-a-Service (MaaS) paradigm.
§ MaaS solutions (e.g., MaaS Global: http://maas.global) aim at arranging
the most suitable transport solution for their customers thanks to the
costs of an effectively integrated offer of different multi-modal means of
transportation.

5.
Our Claim
page
05
§ In the first and last-mile mobility of people and goods, the ASaaS
could be the right concept.
§ ASaaS has the potential of offering various mobility services that:
§ are tailored to the traveller needs and preferences,
§ allow to serve critical areas with minimal new infrastructure and
reducing noise and pollution, and
§ complement the already available (public and private) mobility
services in a city solving the well-known issues of proximity
mobility.

6.
Stakeholders
page
06
§ Municipalities by reducing costs, traffic congestion, emissions, and
energy consumption.
§ The city service providers by offering them an additional client's base,
while granting them the flexibility they need to run their services in a cost
and efficient way.
§ public and private security managers
§ goods delivery companies
§ tourist's office
§ shops and restaurants
§ park sharing companies
§ Etc..
§ The collectivity, facilitating the emergence and the diffusion of
innovative smart mobility solutions, contributing to the reduction of
traffic pressure in cities and supporting the right to mobility also in
disadvantages areas and for disadvantages citizen groups.

7.
ASaaS Platform
page
07
§ IT platform supporting the definition and implementation of a portfolio of
city mobility services that:
§ are tailored/configured to the traveler needs and preferences
§ exploit in a synergistic and collective manner the different already
available mobility services.
§ The mobility services exploit the hardware and software potentialities
of the Autonomous Shuttle that can be configured to be used in
different contexts:
§ City Centers, Hospitals, Private Companies, Stadiums, University
Campus, Ski Centers.
§ For different Goals:
§ Events Management, goods/people delivery, emergency situations,
security management, etc..

8.
ASaaS Platform
page
08

9.
Application Scenarios
page
09
§ Last-Mile Delivery of Goods
§ Optimization of shared city hubs for delivery of goods of different
courier companies, for citizens, bar and restaurants, etc..
§ To combine the provision of services capable of optimizing the
delivery of different kinds of goods in city centers with the smart
transportation of people using the same vehicle, the AS.
§ Tourism / Info Mobility / Geo Marketing
§ A service for helping tourists to have memorable winter and
summers experiences by getting AS to move around the
seaside/mountain town.
§ To improve the tourist experience in cities creating marketing
traction to sponsors and increase brand loyalty and market
penetration.

10.
Application Scenarios
page
010
§ AS as Shared and Integrated Mobility
§ Shared Taxi solution based on AS and covering the last-mile
needs on defined urban routes not supported by taxi or bus
services.
§ A travel planner for all people that also integrate incentive models
and behavior changes enabler (i.e., Gamification).
§ Public/Private Surveillance Management
§ AS for monitoring activities and coordination of responses in
dangerous situations.
§ To collect contextual data about citizen/tourists' habits and
movements with the support of new technologies and motion,
thermal and RFID camera installed onboard.
§ A set of moving AS (i.e., fleets) can collaborate to collectively react
to local and global contingencies – distributed situation
recognition.

11.
Application Scenarios
page
011

12.
Our Vision – The ASaaS Model
page
012

13.
Research Challenges
page
013
§ Simulation and Machine Learning Techniques for Last-Mile Mobility
Planning.
§ A tool that planners can use to optimize the potentially impacts of
these novel technologies.
§ Journeys Tracking Certification via AI and Blockchain-based
techniques.
§ A privacy-preserving journeys certification solution for
sustainable proximity mobility.
§ Journey planning algorithms that must integrate AI and Blockchain
techniques with the objective of tracking and certify users’ journeys
while preserving their privacy.
§ Sustainable Mobility Ecosystem.
§ To minimize the economic, social and environmental cost of the
mobility system.
§ End-users Engagement and Behavioural Change through
Gamification.

14.
Agent-based Modeling
Self-Organization of Collective and
Autonomous Shuttle Fleets
Antonio Bucchiarone, PhD
MoDiS, FBK
bucchiarone@fbk.eu
https://bucchiarone.bitbucket.io/
24 November 2020

15.
The GAMA Environment
page
015
§ GAMA (GIS Agent-based Modeling Architecture) is a modeling and
simulation development environment for building spatially explicit
agent-based simulations.
§ It is available at: https://gama-platform.github.io
§ The GAMA community is active and friendly.

16.
The GAMA Environment
page
016
GAMA Overview:
§ Used in multiple application domains (i.e., transport, urban planning,
epidemiology).
§ Based on a high-level and intuitive agent-based language, GAML, for
modeling agents.
§ GIS and data-driven models: Instantiate agents from any dataset,
including GIS data, and execute large-scale simulations (up to millions
of agents).
§ Declarative user interface: Declare interfaces supporting deep
inspections on agents, user-controlled action panels, multi-layer 2D/3D
displays & agent aspects.

17.
The GAMA Environment
page
017
§ GAML (GAma Modeling Language): is the language used in GAMA,
coded in Java. It is an agent-based language, that provides you the
possibility to build your model with several paradigms of modeling. It
enables users to define agents, as species, whose behaviour is defined
by actions and reflexes.
§ reflexes are called automatically at each time step; actions are
called by an instance of species. Agents can have skills, i.e., built-
in modules with a set of skill-related attributes and actions.
§ GIS and Data-Driven models: GAMA provides you, since its creation,
the possibility to load easily GIS (Geographic Information System). You
can import many data types, such as text, les, CSV, shapefile, OSM
(open street map data), grid, images, SVG, but also 3D les, such as
3DS or OBJ, with their texture.

18.
The GAMA Environment
page
018
Behind the concepts and the operational semantics of GAML, there is a
meta-model. It represents a model with three main categories of abstract
classes representing:
§ Entities. In the model, the representation of entities is based on:
§ Agent: individual entity in the simulation.
§ Population: group of agents with the same characteristics, in terms
of behaviour and structure. It manages the agents that compose it.
§ Space. The two following main classes represent the concept of space
in the model:
§ Geometry: individual entity shape in the simulation.
§ Topology: groupings of agents.
§ Time. Time is represented through the class Scheduler, which is linked
to the Agent class, and the class Scheduling Information, linked to the
Population class.

19.
Simulating the Propagation of a Virus
page
019
Nicolas Troquard – Assistant Professor at the University of Bolzano.
Virus contagion among a population typically follows an exponential
growth. In the event of a pandemic, healthcare systems of a region may
collapse due to the number of patients. Social distancing is a measure of
infection control whose aim is to slow down the spread of a disease
(”flatten the curve") or stop it. Simulating the propagation of a virus:
§ in a non spatially explicit simulation
§ in a spatially explicit simulation in the city of Bozen-Bolzano
We ran three simulations: when 10% of the population is respecting social
distancing, when 60% does, and when 90% does.
https://www.inf.unibz.it/~ntroquard/virusmodel/REPORT/report.html
https://www.inf.unibz.it/~ntroquard/index.htm

20.
Simulating the Propagation of a Virus
page
020
Person's behavior as Finite State Machine.

21.
Self-Organization of Collective and Autonomous Shuttle
Fleets
page
021
The Architecture of the Framework

22.
The Approach
page
022
A decentralized approach for the self-organization of fleets of autonomous
shuttles, is part of the Execution Layer. The computational power for the
grouping of agents and the computation of travel costs are distributed
among the AS.
Main species: people, autonomous shuttle (AS), normal car.
Main steps:
§ Each AS is capable of dynamically finding passengers on the road it is
currently travelling and automatically check whether to offer them a lift,
based on certain conditions.
§ The AS agents compute a path given an origin and a destination on the
road network. This trajectory is composed of a sequence of edges.

23.
The Approach
page
023
§ There is the possibility of finding and adding more passengers on the
road.
§ In the presence of other passengers on the path, the AS will decide
whether or not to offer a lift to them, based on the objectives of the
passengers already on board.
§ Accepting new passengers lead AS to dynamically update their path and
travel costs. The dynamic addition of passengers would be beneficial for
both AS, which can reach their full occupancy, and for the passengers,
who would share the travel costs.

24.
Agents Behaviour as Finite State Machines
page
024

25.
About Travel Costs
page
025
The travel cost is based on the computation of the costs of legs:
For each AS, the travel cost corresponding to the entire path it covered is
calculated by the sum of the costs of the legs in the path:
The cost that each passenger must pay on a given leg for which he/she
was on board:
The total cost for each passenger will be given by the sum of the costlegpass
of each leg he/she has to contribute.

26.
Implementation in GAMA
page
026
https://github.com/Martins83/AutonomousVehicles

27.
Evaluation Results
page
027

28.
Evaluation Results
page
028
Fixed fleet of 50 shuttles in scenarios with varying number of total users,
from 250 to 1000.

29.
Evaluation Results
page
029
Fixed number of 1000 users in scenarios with varying number of AS in the
fleet, from 40 to 70.

30.
AUTOS Laboratory
Leading research and innovation in public
transportation services based on autonomous shuttles
A co-innovation initiative of FBK and NAVYA

31.
AUTOS Lab in a nutshell
Stop #1
§ General scenario :
o Passenger transport services assisted by
volunteer (supervisors as “drivers”)
o Semi-automatic Level 3-4 shuttle of NAVYA
§ We launched a joint-innovation laboratory called
“Autonomous Shuttles-AUTOS”
§ The lab is a technology and scientific collaboration
among FBK, NAVYA e I-MBG (global player in
autonomous shuttles for public transportation)
§ Research and innovation actions on the concept of
Autonomous Shuttle as a Service (ASaaS).
§ Advanced ASaaS model published in the top journal
for automotive engineering IEEE-ITS (Q1)
Stop #2Stop #3Stop #4
Stop #5 Stop #6
31

32.
Lab Objectives
§ Scientific and technological collaboration aimed at digital innovation for
the establishment of a virtual joint laboratory called "Autonomous
Shuttles-AUTOS“.
§ Research activities and innovation projects in the information technology
sector, with specific regard to the field of "Transportation and
Mobility“ for public purposes.
§ The ultimate goal of developing new skills, methodologies,
technologies and platforms that enable digital transformation, as well
as future synergies with public institutions, companies and startups.
§ New models for: Smart mobility, urban mobility, sustainable mobility,
smart transportation, last mile delivery e smart community.
32

33.
Concept: Autonomous Shuttle as a Services (ASaaS)
33

34.
Riva del Garda Scenario: tourist transportation services assisted by volunteer
drivers, on board a level 3-4 semi-automatic shuttle
F1 F2 F3
F6F4
F5
restaurants
beaches
sponsor
34

35.
Some applications cases: worldwide presence of Navya
More information: https://navya.tech/en/autonom-shuttle/applications/
§ Experience of Navya on public roads: ASTRA (Switzerland), FTSA (Finland),
LTA (Singapore), MDDI (Luxembourg), MFK (Liechtenstein), MTES (France), NTC
(Australia), RDW (Netherlands), SAAQ (Canada), SPF M&T (Belgium), STA
(Sweden), TÜV Austria (Austria), TÜV Rheinland & TÜV Hessen (Germany)
35

36.
Selected References
36
• A. Bucchiarone, S. Battisti, A. Marconi, R. Maldacea and D. C. Ponce, "Autonomous Shuttle-as-a-
Service (ASaaS): Challenges, Opportunities, and Social Implications," in IEEE Transactions on
Intelligent Transportation Systems. 2020.
• A. Bucchiarone, M. De Sanctis and N. Bencomo, "Agent-Based Framework for Self-Organization of
Collective and Autonomous Shuttle Fleets" in IEEE Transactions on Intelligent Transportation
Systems, 2020.
• Antonio Bucchiarone. 2019. Collective Adaptation through Multi-Agents Ensembles: The Case of
Smart Urban Mobility. ACM Trans. Auton. Adapt. Syst. 14, 2, Article 6 (December 2019), 28 pages.
• Daniel Furelos-Blanco, Antonio Bucchiarone, and Anders Jonsson. 2018. CARPooL: Collective
Adaptation using concuRrent PLanning. In Proceedings of the 17th International Conference on
Autonomous Agents and MultiAgent Systems (AAMAS '18).
• Antonio Bucchiarone and Antonio Cicchetti. 2018. A Model-Driven Solution to Support Smart
Mobility Planning. In Proceedings of the 21th ACM/IEEE International Conference on Model Driven
Engineering Languages and Systems (MODELS '18).
• Antonio Bucchiarone and Antonio Cicchetti. 2018. Towards an adaptive city journey planner with
MDE. In Proceedings of the 21st ACM/IEEE International Conference on Model Driven Engineering
Languages and Systems: Companion Proceedings (MODELS ‘18).

37.
Contacts:
Antonio Bucchiarone, PhD
Senior Researcher
Digital Society Line, ICT Center
MoDiS Research Unit
Fondazione Bruno Kessler (FBK)
Via Sommarive 18, 38123, Trento
bucchiarone@fbk.eu
https://bucchiarone.bitbucket.io/