2.1 Autonomous versus automated

An ‘autonomous system’ is defined as one that takes actions on its own to achieve goals provided independently by whomever defines them or will benefit from their achievement. Also, the decision-making process is local and offline^{(Reference Helle, Schamai and Strobel13–Reference Figueroa and Walker16)}.

On the other hand, for an ‘automated system’, the action to be performed whenever an unplanned situation occurs is already encoded into the system. A designer must encode all the cases and the corresponding strategies to deal with all such events.

This process occurs externally to the system^{(Reference Frost15)}. An ‘intelligent autonomous system’ simulates human responses using sophisticated mechanisms. Such a system can sense, plan and act independently of any code embedded based on the environment, achieving a goal with no external control^{(Reference Frost15,Reference Beer, Fisk and Rogers17)} .

According to the International Civil Aviation Organization (ICAO) RPAS manual⁽¹⁸⁾, an ‘Autonomous aircraft’ is an unmanned aircraft that does not allow pilot intervention in the management of the flight.

Likewise, ‘autonomous operation’ is defined as that where an unmanned aircraft operates without the possibility of remote pilot intervention in the management of the flight^(18,19) .

With increasing levels of autonomy and its application to unmanned systems, the National Institute of Standards and Technology (NIST) created the Autonomy Levels for Unmanned Systems (ALFUS) workgroup to address this increasing need and compile the work performed in different departments across the United States. The result is shown in Fig. 2, where the level of autonomy is determined by an evaluation of the mission complexity, environmental difficulty and human independence^{(Reference Huang, Pavek, Novak, Albus and Messin1,Reference Huang20,Reference Elbanhawi, Mohamed, Clothier, Palmer, Simic and Watkins21)} .

Figure 2. ALFUS framework, evaluation of different missions. Retrieved from Ref. (Reference Elbanhawi, Mohamed, Clothier, Palmer, Simic and Watkins21).

In this context, a Level of Autonomy (LoA) classification does not yet exist, although it is deemed necessary and is not currently included in aviation safety regulations^{(Reference Clothier, Williams and Perez22)}. The implementation of complex high-level decision-making is already being achieved through the use of neural networks, agent-based architectures and probabilistic reasoning. These technologies remain a challenge for the certification of the level of flight safety in normal aircraft operations^{(Reference Clothier, Williams and Perez22)}. Figure 2 provides more detail.

Another industry approach was defined and applied to autonomous road vehicles by the Society of Automobile Engineers (SAE). They defined five levels of autonomous driving, as presented in the table in Fig. 3, where a correlation matrix of the levels of autonomy and the activities performed by the system or driver is presented. For lower levels of autonomy, more intervention from the driver is required. As the level of autonomy is increased, more activities are performed by the system, thus diminishing the driver’s responsibility in taking control and decisions. At level 5 (full automation), no human intervention is required in the different stages of driving^{(Reference Litman23)}.

Figure 3. SAE driving levels of autonomy – retrieved from SAE International.

The automation of a system can be evaluated and classified at many levels regarding its performance^{(Reference Parasuraman, Sheridan and Wickens24)}. Another method of evaluation that is commonly used is the levels of automation of decision and action selection, which has ten levels, where level 1 involves no computer assistance while level 10 requires no interaction with humans with decisions being made only by the system^{(Reference Parasuraman, Sheridan and Wickens24)}.

Also in the aeronautical domain, automation has been studied over recent years, to reduce pilot activities and thus the need for two pilots on board. Aircraft operations are now more reliant than ever on automation to ensure safety levels with the introduction of new functionalities, responsibilities and effective human–autonomy teaming^{(Reference Bailey, Kramer, Kennedy, Stephens and Etherington25)}.

There is a general feeling that “the greatest obstacle to the development of a civil, single-pilot aircraft is not the technology per se but applying the technology and developing the automation and user interfaces”, enabling near-zero crew activity to become a reality^{(Reference Bailey, Kramer, Kennedy, Stephens and Etherington25)}.

General aviation is still far away from fully autonomous piloting, although the introduction of some degree of automation has brought advantages and disadvantages related to pilot skills and awareness. Some advantages are described in the work developed by the European Aviation Safety Agency (EASA) Internal Group on Personnel Training⁽²⁶⁾ and can be summarised as follows:

• – Flight safety improvement, i.e. envelope protection, velocity limitation, etc.

• – Improved technical reliability

• – Aircraft control (from the engine to navigation accuracy and improved aerodynamic performance)

• – Improved flight path control

• – Increased navigation awareness

• – System monitoring and diagnostic assistance

• – Automation relives pilot from repetitive tasks, reducing workload

In the same way, disadvantages can also arise from automation in the form of:

• – Reduced capability for interaction among crew

• – Gaps in training due to different levels of flight automation between aircraft

• – Reduced task/alert prioritisation capability

• – Reduced standardisation and workload management

• – Erosion of manual flying skills of pilots

2.2 Urban areas with high population density

In terms of the operational environment, ICAO annex 11 divides airspace systems into classes A–G, reflecting their operational requirements. States can choose how to separate and classify their own airspace. Class A, B and C airspace is mainly used by commercial transport aircraft. Classes C, D and E are used by general aviation that normally operates out of small airports. Class G corresponds to very low heights and uncontrolled airspace outside of controlled airports. Class F airspace is designated as ‘uncontrolled’, although ATC clearances and separation can be provided, mainly when entering and exiting the area. Figure 4 depicts each of these airspace classes as used in the United States.

Figure 4. Airspace classes, retrieved from Ref. (28).

In addition to the abovementioned classes, Atkins^{(Reference Atkins27)} proposes a subdivision of airspace class G to include class U from the surface to 500ft (ca. 150m) above ground level.

At the moment, several projects are being developed to enable flight in class G airspace, namely integrating UAS operation into an urban environment, thus creating the need for an Unmanned Traffic Management (UTM) system. One bottleneck identified in increasing the number of aircraft in class G airspace is the ATC workload, so it is envisaged that an architecture similar to UTM will be adopted by UAM^{(Reference Thipphavong29)}.

The UTM concept defined by the Global UTM Association, a Swiss non-profit organisation, states the following:

Low^{(Reference Huat-Low31)} presented his vision for management of the Singapore urban airspace for UAS operation. He proposes that the city be mapped and defines several strategies to ensure effective unmanned aircraft operations. These strategies for flight in urban airspace are presented in Fig. 5, where several constraints are imposed, namely a safe distance between aircraft, limitation of entering/leaving operational areas with the introduction of geofencing, route delimitation and pre-designated landing and take-off areas⁽³²⁾.

Figure 5. NTU proposed concept for air traffic management solutions for unmanned aircraft systems. Retrieved from Ref. (32).

Pathiyil et al.^{(Reference Pathiyil, Low, Soon and Mao33)} also presented some solutions to overcome UAS traffic management challenges with the creation of specific lanes/tunnels, the use of existing infrastructures and the application of geofencing technology with vertical and horizontal separation.

The operation of unmanned aircraft in an urban environment leads to challenges such as the need to prevent collisions between unmanned vehicles or even with manned aircraft or other flying objects and buildings in the urban skyline.

This context makes it necessary to address the airspace capacity, complexity, safety and efficiency of UAS operations in a holistic manner^{(Reference Pathiyil, Low, Soon and Mao33)}. The inclusion of autonomous aircraft with passengers on board will increase the complexity and create more safety concerns related to these areas.

Holden and Goel^{(Reference Holden and Goel34)} presented Uber’s view on the business model with a feasibility study for an eVTOL aircraft service and cost analysis for consumers. Given the existing technical knowledge, they conclude that it is possible to achieve safe personal transportation within an urban area. They also conclude that, for a certain scale of business, a VTOL service is possible at a reasonable cost, eventually leading to mass operation.

To enable the use of autonomous aircraft in urban areas, new infrastructure/hub concepts are identified:

• – ‘Vertiports’ (VTOL hubs with multiple take-off and landing pads, as well as charging infrastructure)

• – ‘Vertistops’ (a single VTOL pad with minimal infrastructure)

Regarding the regulation of the construction and operation of heliports, guidelines exist from civil airworthiness authorities based on ICAO annex 14 volume II. The approval requirements for the design of US heliports can be found in FAA advisory circular (AC) 150/5390-2C. The EASA is still developing rules under Notice of Proposed Amendment 2017–14, to be called CS-HPT-DSN^(19,35) .

In the aforementioned regulations, requirements are provided regarding operational areas, taxiways and taxi routes, helicopter parking, obstacle limitation and other requirements.

One aspect presently not addressed in the available literature is autonomous aircraft operating under a UTM system and using a heliport approved for Visual Flight Rules (VFR) and that can be used by other aircraft operating with a common ATM system.

The US air traffic management system has recently seen new developments with the introduction of technologies such as automatic dependent surveillance-broadcast, performance-based operations, weather integration and network data communications through NextGen programme of the National Aeronautics and Space Administration (NASA)^{(Reference Sridhar and Kopardekar36)}.

In Europe, civil aviation safety is ensured by Basic Regulation (EC) N0 216/2008 of 20/02/2008 on common rules that define the EASA by ‘force of law’ as the entity to supervise and interact with National Aviation Agencies (NAAs).

With regard to product certification, i.e., type certificates issued by the EASA, several documents are available with different levels of requirements to demonstrate and gather evidence on aircraft safety and the ability to fly safely.

These certification specifications address the applicable product system to be certified and mainly the number of passengers, Maximum Take-Off Weight (MTOW) and type of operation to be performed (performance, VFR, Instrument Flight Rules (IFR), etc.).

Additionally, the said certification specifications include subparts with requirements to certify onboard equipment (based on FAR/CS 25.1309). As a complement to certification specifications, acceptance means of compliance apply to them, providing information on procedures and activities to demonstrate compliance to a safety level.

Finally, certification specifications consider the use of supporting documentation, which is also referenced and available through standardisation groups to ensure a certain level of analysis that substantiates safe performance of the system, where the safety assessment process guideline is presented in SAE’s ARP4754 international standard⁽³⁷⁾.

In Europe, the current safety review process is based on the CS 25.1309 certification requirement that calls for Aerospace Recommended Practices (ARPs) developed by ASTM, namely ARP4754 and ARP4761.

Based on this discussion, it becomes necessary to consider different systems architectures that will enable autonomous flight operation, ensuring the safety of passengers as well as aircraft flying in the vicinity of autonomous aircraft and having at least an Equivalent Level of Safety (ELOS) to that of manned aircraft.

Table 1 presents the relationship between the failure condition and the allowable probability of the event per flight hour for manned aircraft, where a catastrophic event should have an extremely low probability of <10^–9.

Table 1 Relation of probability with severity of failure condition. Retrieved from Ref. (39)

Note: A numerical probability range is provided as a reference. The applicant is not required to perform a quantitative analysis, nor substantiate by such an analysis that this numerical criterion has been met for minor failure conditions. Current transport category aircraft products are regarded as meeting this standard simply by using current commonly accepted industry practice.

The implementation of more technology implies a greater demand for software as well as extensive compliance with industry standards such as DO-178C^{(Reference Torens, Adolf and Goormann38)}.

It can thus be stated that the type of operation, systems architecture, environment and interaction with other aircraft will always impact on safety and should be addressed in certification specifications.

2.3 Operation of RPAS in European airspace

Another aspect is the operation of RPAS in European airspace. According to Regulation 2018/1139 of the European Parliament and of the Council of 4 July 2018, operations should be subject to rules that are proportional to the risk of the particular operation or type of operations regardless of aircraft mass. National authorities have the responsibility to implement a risk-based approach while ensuring an adequate level of safety^(40,41) .

EASA opinion 01/2018 published on 6 February 2018, proposing a new European regulation for UAS operations, laid down regulations regarding the operation of unmanned aircraft for open and specific categories as well as for minor- to higher-risk operations, respectively.

These regulations were approved in June 2019 by Commission Delegated Regulation (EU) 2019/945 and Commission Implementing Regulation (EU) 2019/947. These include a 2-year transition period. For risk-based assessment of operations, EASA adopted the Specific Operation Risk Assessment methodology developed by the Joint Authorities for the Rulemaking of Unmanned Systems (JARUS) group.

Operations in the RPAS open and specific categories are under the responsibility of national civil authorities, with some differences between countries.

In this context, UAS operations are approved after a systemic operation review, which includes a risk assessment by the UAS operator.

As standard scenarios are still under development and not yet issued, operations are approved on a case-by-case basis, based on:

• – An operational plan

• – An emergency and contingency plan

• – Detailed specifications of the UAS (whichcan include detailed information on communications systems and procedures, RPAS configuration, flight limits, etc.)

• – A Specific Operation Risk Assessment (SORA) for the proposed operation

Also, the organisation proposing to conduct the operation must supply evidence of the pilots’ proficiency and sufficient knowledge to ensure flight safety as stated in the required documentation.

EASA has still to approve regulations to create RPAS training organisations that may issue RPAS certification for pilots and technicians mainly for the operation of specific aircraft. The development of regulations for certified operations by the EASA is still underway, with no planned publishing date.

One objective of the present work is to analyse strategies and technologies used in UAS developments to enable an assessment of whether the current approach can be extrapolated to urban air mobility.

UAS operators use a risk-based approach to achieve operational approval. If the risk is higher, a certification process is required. Unmanned aircraft operators use SORA to perform a risk-based approach using kinetic energy and operational risk analysis, i.e. depending on size and maximum velocity. Operational risk is assessed through a qualitative perception of risk to people and infrastructure that are overflown by the RPAS.

2.4 Aircraft systems

Aircraft designs and configurations have evolved in recent years as a consequence of UAM needs. Currently, more than 240 vehicles are being developed worldwide⁽⁴²⁾. Major aircraft manufacturers such as Airbus, Boeing, Embraer and Bell and even automotive industry manufacturers (e.g. Audi and Nissan) or start-up companies (e.g. Volocopter and Lilium) are currently developing concepts for aircraft that can operate in urban environments.

eVTOL aircraft can assume multiple configurations, and it is difficult to track the developments of the many start-ups and other more well-known organisations^{(Reference Bacchini and Cestino43)}. One source to keep track of such developments is the Electric VTOL News ⁽⁴²⁾, where more than 170 aircraft are classified into several categories:

• – Vectored thrust

• – Lift + cruise

• – Wingless (multicopter)

• – Hover bikes/devices

• – Electrical rotorcraft

Clarke et al.^{(Reference Clarke, Smart, Botero, Maier and Alonso44)} presented a comparison of the performance based on a mission profile, leading to a new aircraft development methodology that demonstrates the dependence between the configuration and mission objective.

Silva et al.^{(Reference Silva, Johnson, Solis, Patterson and Antcliff45)} analysed three aircraft configurations: quadrotor, side-by-side and lift + cruise. All these configurations generated issues and several areas for future research, including:

• – Tools to develop and analyse aircraft performance

• – Datasets for validation

• – Propulsion technologies and batteries

• – Edgewise fixed-pitch rotors

Aircraft configurations are related to the mission objective and specific performance requirements. This statement is based on the thrust typology per aircraft mass, as demonstrated in Fig. 6. This figure reveals that the vectored thrust typology is the solution most often adopted and implemented in the mass range from 400 to more than 1,500kg, although other typologies can also be adopted in this range.

Figure 6. Thrust typology per aircraft weight. Retrieved from Ref. (42).

2.5 A review of UAM operational concepts

Several authors have envisioned operational concepts for UAM. NASA has been one of the main sponsors of UAM. They are working on a main concept to implement high-density operations using a small number of vertiports interconnected with a main hub and managed by a UAM operator. They are also planning in a next phase to develop mature UAM operations in a network with multiple hub locations, potentially with several vehicles and operations^{(Reference Thipphavong29)}.

The following types of operation are foreseen:

• Private air transport

• Personal scheduled transportation

• Personal unscheduled transportation (on-demand mobility)

• Commercial scheduled transportation

Greenfeld^{(Reference Greenfeld46)} proposed several scenarios for different types of operations in an urban environment using a communication system for Command and Control (C2) with a remote pilot in command, considering all structural obstacles and radiofrequency noise/threats.

Prevot et al.^{(Reference Prevot, Rios, Kopardekar, Robinson, Johnson and Jung47)} proposed a UTM concept for the operations and architecture with information flow for controlling the operations of small, unmanned aircraft in uncontrolled airspace (class G). No interaction with any other airspace class or ATM system was envisaged by those authors.

Various strategies for airspace integration were proposed by Lascara et al.^{(Reference Lascara, Lacher, DeGarmo, Maroney, Niles and Vempati2)}, including:

• – Augmented visual flight rules

• – Dynamic delegated corridors

• – Automated decision support services

• – Performance-based operations

An alternative concept to UTM is that of Remote Operations Centres (ROCs), presented in Fig. 7, which would interface with ATC and be responsible for traffic management and ensuring safe operations of aircraft flying in the area^{(Reference Nneji, Cummings, Stimpson and Goodrich48)}. This concept is halfway between a fully autonomous system and one with several levels of human intervention.

Figure 7. ROC system boundary. Retrieved from Ref. (Reference Nneji, Cummings, Stimpson and Goodrich48).

Although international operational rules and procedures for the flight of autonomous aircraft in urban areas are yet to be developed, OEMs are proposing requirements to be considered during ongoing aircraft development phases via guidelines available from other aviation areas such as light aircrafts and VTOL aircraft, with some simplification.

Concerning the range of operations, Fig. 8 presents a comparison of travel distance versus time for a car and a VTOL aircraft. Bacchini and Cestino^{(Reference Bacchini and Cestino43)} present reference missions related to distance as follows:

• – 7km, defined as an urban mission

• – 30km, named an extra-urban mission

• – 100km, defined as a long-distance mission

Figure 8. Comparison of 40min travel between car and eVTOL aircraft, retrieved from Ref. (Reference Scalabrin49).

Operational concepts that enable UAM must define and establish all the boundaries related to safe aircraft operation in an urban environment and cover at least the following topics:

• ∘ Operational environment

• ∘ Operation constraints

• ∘ Operational use cases

• ∘ Pre-flight phase

• ∘ Flight phase

• ∘ Post-flight phase and ground operations

• ∘ Airworthiness

The resolution of these topics will enable the development of requirements to address all the shortcomings and issues for this new type of operation. These topics are developed in the next section.

Having identified the relevant aspects of the operation of UAM in urban areas, the next section addresses the operational concepts that will create reference conditions to allow the establishment of a high-level functional architecture describing how this type of systems can be operated.

3.1 The approach

Although the term ‘Concept of Operations’ (ConOps) is mostly used in the defence sector, this methodology is fully applicable to capture the requirements of UAM operation with unmanned aircraft, given the nature and type of information contained in such documents.

Despite being a generic document, the ConOps addresses the aspects that an operator needs to consider, mainly from a mission standpoint or eventually from a lifecycle perspective, covering inter alia technical, regulations and support aspects.

Likewise, using the ConOps reference approach to capture UAM operations, the conditions contained therein enable the identification of several specific aspects.

This section provides a compilation of the various operational aspects that are considered to enable UAM operations with unmanned aircraft – either remotely piloted or with autonomous operation capability.

Given the nature of this section, we do not attempt to establish legal, technical or operational requirements, but rather a set of conditions to act as a reference for operational concepts.

In brief, the strategy of this section is to derive a neutral, global (not for a specific operator), realistic and plausible ConOps for UAM operations using unmanned aircraft. This ConOps aims to capture the specific aspects that can only be dictated by aeronautical regulations or even by imposing the existence of specific aircraft systems that will provide certain operational features, e.g. collision avoidance, terrain alerts, automatic landing, etc.

In this respect, the considered ConOps applies to urban areas with a high density of buildings, traffic and people, essentially providing the conditions for the associated business models.

This ConOps is aligned with the On-Demand Mobility (ODM) business model using an autonomous aircraft with VTOL capability.

In this perspective, the main objectives of the ConOps are to:

• – Establish the relevant stakeholders

• – Define the operational environment strategies

• – Describe the operational cases

• – Outline the operational scenarios

• – Delimit the operational conditions

3.2 Stakeholders

The ConOps definition starts with the identification of the stakeholders. The stakeholders for the UAM are considered to be:

• – Authorities and regulators

• – Airspace management service providers

• – UAM service providers (operators)

• – Infrastructure providers

• – OEMs

• – UAM users

The definition of the group of stakeholders is critical because each entity has a significant impact on the development of UAM.

Analysis of each stakeholder is an effective method to understand its contribution to the UAM process⁽⁵⁰⁾. A structured stakeholder analysis along with the interaction of stakeholders for the different phases and the modelling process is proposed in Table 2.

Table 2 Stakeholder impact analysis in the UAM process definition

3.3 UAM unmanned aircraft missions: outlook

To act as a reference characterising aircraft systems and airspace management capability, a typology for missions for unmanned aircraft for UAM to be requested by potential users is presented below:

• – Range: 100km

• – Reserve: At least 20min in hovering condition

• – Maximum operating altitude: 6,000ft above MSL

• – Minimum operating altitude: limited to 300ft above MSL

• – Distance to buildings: dependent upon local noise restrictions

• – Cruise velocity: 150km

The following operating segments are currently considered as a reference for operation:

• – Taxi and flight preparation: 5min

• – Climb: 8min

  • ∘ Vertical: 3min

  • ∘ Transition to forward flight: 5min

• – Cruise distance: according to trip range

• – Descent: 8min

  • ∘ To hover condition: 5min

  • ∘ Vertical landing: 3min

• – Land and power-off: 5min

• – Abort and diversion: 20min

Figure 9 graphically summarises each of the above ConOps conditions for UAM.

Also, the existing 14 CFR 91.151 regulation from the FAA requires a reserve of 20min of flight time at cruise speed for VFR rotorcraft^{(Reference Finger, Gtten, Braun and Bil51)}, while the EASA Part-CAT operating procedures for helicopters state the same 20min flight time reserve for VFR or 30min for IFR flights⁽⁵²⁾.

Figure 9. UAM operational concept, segments (developed by the author).

Although the estimated total flight time for this type of operation is less than 40min, a reserve of 20min is proposed for cruise and loiter from an aborted landing to a new landing site. This approach is very conservative as it considers a 50% energy reserve for operation under normal conditions. In the future, this should be further elaborated to confirm its effect on safety and whether this percentage can be reduced.

3.4 UAM meta-system: outline

UAM operations are dependent on the capability of future UAM unmanned aircraft to use urban airspace while ensuring safe operation for people in the air and on the ground.

Future UAM unmanned aircraft can be decomposed into several groups that will communicate and transfer data between them. Figure 10 details the data flow foreseen between the various components of the UAM meta-systemFootnote ¹.

Figure 10. UAM meta-system interaction (developed by the author).

This outline of the UAM meta-system is of fundamental importance to the operational concept because it dictates how the operators will develop and carry out each UAM mission.

3.5 Operations

The various elements of the operational features essential for the formulation of the UAM ConOps for unmanned aircraft are described below:

3.6 Regulatory framework

Despite the obvious inability to anticipate what type of regulatory elements will be established by the various authorities granting authorisation to perform UAM unmanned aircraft operations, it is unavoidable to assume that the applicable ConOps shall be constructed taking into account those specific regulations, which tend to address at least the following aspects:

• – The physical and functional configurations of the authorised unmanned aircraft

• – Operational routes and alternate landing sites after an emergency situation

• – Security and safety procedures

• – Requirements in terms of persons involved in UAM unmanned aircraft operations

The next paragraphs of this section provide those regulatory elements. Based on the state of the art, it is anticipated that these will be enforced by the aeronautical authorities and thus must be considered in the UAM unmanned aircraft operator ConOps.