More than 50% of the world’s population lives in cities [1], and it is expected than 7 out of 10 people will do in 2050 [3]. Cities have to evolve fast to cope with this growth and adapt to human needs [2]. The CS Group believes that communities showing live-work symmetry, walkable amenities and local production, three ideas gathered around the concept of a 15 minutes walkable city, will be better positioned to be high-performing and efficiently address the complex growth.
New designs of mobility, urban planning and architecture, implementing new technologies and fostered by new public policies are the current needs in urban areas and objective of study for the CS Group. With this master’s thesis, various contributions have been made towards this model of a 15 minutes walkable city:
First, the needs to transport people, goods and perform utility services in this new cities has been addressed. A framework for future mobility based on collaboration in vehicle fleets has been outlined. This framework follows the current trends and it is inspired by how nature has evolved to solve similar challenges. This new framework brings: The necessity of new ultralight vehicles, new infrastructure and new operation. Contributions have been made in these three branches to build towards this future mobility.
Second, urban planning processes need to evolve to be more agile and include non-experienced community members in the decision process. The CS Group has worked for years in prototyping tools to improve the urban planning process. A novel contribution has been made to this tools, leveraging the machine computational power to assist urban planners in creating and optimizing their models. Besides, the CS Group is now working in a new concept of urbanism, where cities will grow vertically. This is the most promising urban planning solution for the growth and complexity challenges humanity is experiencing. Contributions have been made to this vertical urbanism, impacting two key steps of the process.
5.1 A new urban mobility in the 15 minutes walkable city
5.1.1 Future mobility framework - a bio-inspired proposal for a collaborative sys- tem of multi-functional autonomous vehicles
In the current trend towards electrification, autonomy, and sharing, little attention is being paid to what will come after these trends are settled, specially from the perspective of interactions between vehicles and how these interactions might affect the system-level behavior. Mobility is an increasingly complex system, and, in nature, there are various examples of complex systems with similarities to mobility that have evolved to be collaborative. Research in swarm robotics has demonstrated the potential benefits of translating these natural behaviors into artificial systems. Moreover, research in vehicle platooning has demonstrated that coordination between
vehicles can lead to more efficient traffic, reduced energy consumption, and increased safety.
The outlined framework proposes a future that combines, expands, and translates ideas from nature, swarm robotics, and vehicle platooning, proposing a future mobility based on three key ingredients: system behavior, multi-functionality and collaboration. In addition to inheriting the benefits of vehicle platooning and swarms, this proposal can radically transform the way cities address the three main mobility needs: the movement of people, movement of goods, and utility services. A pre-print has been published outlining this framework [18]
This framework points towards a future in which mobility would work more similarly to how natural systems behave. Since natural processes have been tailored by millions of years of evolution, natural systems are more efficient, sustainable, and resilient, which are characteristics that future transportation systems would ideally have.
5.1.2 Developing the driverless skills for a lightweight mobility prototype
With this first iteration, a proof-of-concept algorithm was developed, showcasing how a light- weight mobility vehicle such as the MIT Autonomous bicycle can circulate inside a lane. The code, developed from Madrid (Spain) during the pandemic, was successfully tested at MIT (US), where the MIT Autonomous bicycle was able to follow a lane during the test. Moreover, extra equipment added to the bike was low-cost, composed by just a 1080p webcam to perform the vision tasks.
From this point, development was started on a second version for the lane following algorithm.
In parallel, research was conducted on how to make the city more intelligent. This research included using smart systems to guide this ultra lightweight mobility vehicles, in order to lower their cost by freeing them of complex systems as Lidar. For a fast proof of concept of those ideas, the idea of using a simulator was explored.
Engineering simulation have been performed for decades and have shown to be key in the devel- opment of cutting-edge technologies and new products. According to ANSYS (a major developer and supplier of simulation toolsets), thousands of virtual tests can be completed withing the time and budget available for a single physical test, thus greatly accelerating technology de- velopment [284]. As they mention, simulation gives a faster time-to-market, reduced costs and enhanced product quality.
In the case of autonomous vehicles, simulation helps to build more performing and safe driverless vehicles. Without simulators, driverless vehicles would drive millions or even billions of kilomet- ers to achieve the same reliability. With this idea of working on a simulator, research was made on state-of-the-art simulators that could allow us see how smart systems such as beacons could be implemented in cities to guide bicycles. Some simulators such as AirSim [285], Apollo [286], Autoware [287], Carla [288] and Gazebo [289] were analyzed.
This analysis showed that more resources where needed, notably a bigger team, to achieve a great impact with those simulators. Thus, the decision was to prioritize the infrastructure and operation contributions towards the future urban mobility.
5.1.3 A system for dynamic V2V power-sharing in collaborative fleets of autonom- ous vehicles - one step towards the proposed framework
In the current trends towards electrification, autonomy, and vehicle-sharing, vehicle charging is one of the main issues preventing vehicles from populating the roads. Current solutions are either inconvenient or require expensive and complex infrastructure. In the proposed collaborative future mobility, vehicles will share battery with other vehicles while moving, and vehicles will access energy anywhere and anytime.
Such approach reduces infrastructure costs and provides great flexibility, eliminating the problem of accessing stations. It optimizes the vehicles usage at a system level, with less time and kilometers traveled related to charging. Since vehicles can charge while moving, vehicle downtime is reduced, consequently increasing vehicle usage.
Since there was no technical solution for this application, this master’s thesis brings a connection that allows vehicles to share battery while moving. The proposed connection combines the usage of inductive charging together with an electromagnet.
This connection is easy to connect-disconnect, tunable, durable, bidirectional, safe, and efficient.
A proof of concept prototype was developed. The prototype was composed by a fully functional vehicle, charged and able to move, and a discharged vehicle. The fully functional vehicle ap- proaches and tows the discharged vehicle, validating that the proposed connection allows for easy attaching/detaching and wirelessly sharing energy while moving. An article has also been published outlining this solution [17].
5.1.4 Agent-based model to test a new rebalancing scenario based on stimergy A future in which multi-functional vehicles will collaborate and work as a system, opens mul- tiple questions about how such system can operate. In order to address those questions, an agent-based simulation has been built, able to replicate those mobility systems and show their characteristics and potential benefits.
As a first iteration, the rebalancing problem for shared electric vehicles has been chosen. This proposal presents a decentralized alternative for vehicle rebalancing in fleets of shared autonom- ous micro-mobility vehicles based on bio-inspired foraging behavior. The proposed approach has been modeled in a realistic multi-layer agent-based simulation. The results indicate that the system can effectively redistribute the bicycles in a self-organized manner, significantly reducing the wait times compared to not rebalancing the vehicles. The results also show that wait times are reduced over rebalancing vehicles randomly; however, results indicate that the evaporation and exploitation rates need to be appropriately tuned to get the desired behavior.
Finally, when designing these systems, it should be kept in mind that some parameterizations, especially those with high wandering speeds and large fleet sizes, can significantly increase the total distance traveled.
5.2 A new urban planning for the 15 minutes walkable city
5.2.1 2D genetic city - A human-machine interaction with human intelligence aug- mentation tool for urban planning
There is an urgent need of understanding how cities “work”, as well as the well-being of their inhabitants and the possible impact of various urban interventions. The MIT City Science Group has been working for years in the City Scope tool, which allows rapid prototyping and evaluation of urban interventions. Besides, the City Scope focus on democratizing the decision making process so that not only a few stakeholders build consensus for new urban interventions but non-experienced community people make part of this process.
However, it is difficult for non-experienced stakeholders to propose solutions with such amount of inter-dependencies among elements and with multiple objectives (e.g. economic, environmental, health) at the same time. Furthermore, this is becoming a challenge even for experienced with the increasing complexities embedded in urban scenarios.
Thus, a human-machine interaction with human-intelligence augmentation tool has been pro- posed, to inspire stakeholders in building data-driven proposals, complying with their objectives and which they understand and agree with. A proof of concept has been developed with a custom built genetic algorithm. This has served as a first iteration to gather feedback for the human- machine interaction part and to understand how multi-objective evolutionary algorithms work.
These algorithms are very promising in solving nonlinear complex multi-objective problems.
Testing the tool with both experienced and non-experienced users has shown the potential it has for both inspiring and guiding users towards a desired solution complying with their objectives.
These users have reported positive feedback about the tool, together with possible ideas for future iterations.
Thus, the tool has proven how interesting it is to develop a more solid and complete version, with state-of-the-art multi-objective evolutionary algorithms. These algorithms will approach the Pareto front faster and with more robustness, unlocking the potential of the tool to be successfully applied in very different urban contexts.
5.2.2 The new City Scope platform to design and evaluate vertical urban scenarios The population growth and its migration to cities is putting pressure on land consumption, as it is expected to add 1.2 millionkm2to new urban built areas [3]. This is critical as well for natural resources, and it requires new forms of urban planning. Although urban morphologies need to be more compact to attain with this density objective, there is a current situation where verticality has been implemented in single buildings (centralization), while the urban area continues to grow laterally (decentralization) [2].
Vertical urbanism comes to allocate the population growth and its migration to cities, without increasing the urban sprawl and in a sustainable manner. This is achieved increasing density while addressing big challenges such as public health, global warming and equity. Moreover, vertical urbanism can bring benefits such as increased productivity, walkable amenities or spaces saving [23]. Nevertheless, vertical urbanism also brings new challenges such as lack of open spaces or greeneries, lack of privacy or noise pollution [23] which need to be adressed as well.
In the CS Group there is an ongoing effort to understand how hyper-performing and resilient
communities can grow in a vertical city, as it will be the only way to comply with the vastly increasing density needed. Thus, the CS Group is working in a tool that allows to both design and evaluate vertical urban scenarios and possible urban interventions, a reinvention of the City Scope platform. This tool will be useful to perform quick iterations and design urban areas that address great challenges humans are facing (e.g., public health, global warming, equity).
The proposed tool, which can be applied to any urban area, is composed by various parts: (1) a study of the amenities needed, (2) the creation of the CAD file outlining a design that contains those amenities, (3) a link & nodes structure built to analyze that CAD file, (4) a proximity analysis run on the data structure created, (5) a generative tool to inspire urban planners in the creation and optimization of their models, and (6) a visualization of the results of both the proximity analysis and the generative tool.
This master’s thesis has contributed with two of those parts, (3) and (5):
(3) For the links & nodes structure, a pipeline has been built in Python. As an input, it reads a design in a CAD file which contains those amenities defined as necessary when studying the urban area. Then, a data structure with all the amenities and their connections (distances) is generated. With this data structure, the proximity analysis has been performed, showing the strengths and weaknesses of the proposed design. Moreover, the analysis proves the right functionning of the data structure creation.
(5) This generative tool constitutes the second version of the one developed as well in this master’s thesis. This new version of a generative tool needs of more land uses and objectives, being more representative of reality. This adds more complexity, for which state-of-the-art algorithms are needed, and need to be fine-tuned to give meaningful results. The structure of the problem has been defined, and a Python library has been proposed, outlining the steps to proceed.