SEEDPOD

Simulation Environment for the Evaluation of Drone Policies and Optimal Deployment rules

An illustrative diagram of the initial SEEDPOD vision

1         Context

Within both the UK as well as internationally, there are currently many initiatives aimed at expanding the use of drones as well as other forms of un-piloted aircraft such as urban air mobility aircraft. However, there are many imponderables regarding the rules and policies that should govern their use. Many physical trials and case studies are being carried out and being planned to prove the technology.

The SEEDPOD initiative aims to augment these trials by helping to guide policy makers and regulators by providing a detailed, realistic and wide-ranging simulation capability that can help quantify the risks and impact of a significant expansion of drones and other aerial robots in the future.

2         Objective

The goal of SEEDPOD will be to develop a software tool that can be made freely available to users in order to analyse the outcome of deploying a heterogeneous collection of platforms undertaking a diverse number of missions in a geographic area.

Its primary purpose will be to evaluate drone policies and rules in a given region. For example, city and county councils, as well as other government authorities, will need to determine such things as overflight rules, flight corridor heights, noise limits and permission criteria.

3         Review

Before work takes place on this software tool a comprehensive review of existing research work, commercial and open source tools and academic papers will be undertaken. An initial review has already been carried out and relevant work listed in appendix A

4         Example Usage

Figure 1 – Example benchmark region: Solent showing “snapshot” of transponder carrying manned aircraft.

The purpose of this tool is to analyse the impact of decision variables such as drone policies and to understand their implications. In order to do this, a hypothetical Unrestricted Baseline (UB) will be established as explained in more detail below. The UB would take a specific geographic region (see Solent Region image below taken from Flightradar24.com) and create a future projection of drone services and businesses and their associated flight patterns.

An example UB based in the Solent Region would for example model the following drone applications. The table below gives an estimate of the number of concurrent drone flights that might take place in this region in the “near future”.

Figure 2 – Estimate of types, roles and numbers of future Unmanned Aircraft within the near future for use in a benchmark SEEDPOD simulation with the Solent Region.

4.1        Unrestricted Baseline (UB)

The idea behind this baseline would be to show the effect of the introduction of drone rules and policies compared with some semi-arbitrary benchmark. The UB might, for example, assume that a highly permissive environment is allowed with few restrictions on drone usage.  Although this benchmark is essentially unrealistic its purpose is to provide a comparison point. It is required in order to demonstrate that rules and policies are imposed for a clear purpose and lead to measurable net overall benefits. The UB could for example have;

  • No restrictions on altitude thus allowing drones to fly both high and low
  • No restrictions on where drones can fly
  • “Laissez-faire” restrictions on operations
  • No UTM
  • No airworthiness rules

4.2        Metrics

For each simulation run, a number of metrics would be generated as outputs. These will include;

4.2.1        Aerial collision risk

The simulation would measure the frequency and severity of “air-prox” incidents both between drones and drones and manned aviation.

4.2.2        Ground risk

The simulation will calculate the kinetic energy effect of failed drones dropping onto areas of varying population density (therefore threat to life or cause of injuries) and infrastructure/ property damage probabilities.   

4.2.3        Service levels

This would be a measure of how well the services levels relating to the drone operations listed in Figure 2 are maintained.

4.2.4        Cost comparisons

This metric would show the relative cost of operations as rules and regulations are introduced. For example, a restricted area might require a drone to fly around this obstruction rather than a direct flight. The simulation will be required to capture these extra costs both in terms of flight times and delay to services and other associated costs.

4.2.5        Environmental impact

The simulation should provide a way of showing the noise impact of operations in some suitable comparative metric including violation of constraints.

Similarly, provision should be made for comparative studies showing the emission implications of various propulsion technologies.

5         Structure

The tool will allow scenarios to be simulated and for output statistics to be computed.

It will consist of the following functional units:

5.1.1        Geographical toolbox

This allows users to;

Define the geographical boundaries of the simulation

Identify features that will be needed in order to run the analysis such as roads, land categories, topology, drone airspace classifications, public and private infrastructure etc

This toolbox should allow a wide variety of GIS and other open source and relevant commercial map sources to be accessed to define simulation scenarios.

For Urban operations “city-scape” data will be required detailing buildings as 3D objects as well as other important features

5.1.2        Platform database

This will hold a database of technical data for a range of drones and other un-piloted aircraft including;

5.1.2.1         Performance data (payload/range)

This should allow the simulated platform to respect and authentically model the performance characteristics of the drone/ unpiloted aircraft with sufficient fidelity. This would include airspeed, weather tolerance, payload capabilities, endurance and reserves.

5.1.2.2         Noise models

Noise models will be required particularly for low altitude and urban operations where noise is likely to be a particular sensitivity. There may have to be several levels of fidelity of these models with the most basic representation being a noise limit “sphere”. More sophisticated models may be required for example for large VTOL platforms that may radiate considerably more noise on take off and landing for example.

5.1.2.3         Reliability models

Each platform should have a representation of the likely reliability of the overall system and the consequences of various failure modes including sensors, flight control, conspicuity and communications.

5.1.2.4         Autonomy capabilities

Platforms may have a range of levels of autonomy. At the lowest level, the platform might simply be remotely piloted which requires a secure command link and reliable situational awareness from the control point. Such a drone will require a human operator for every instance of the drone. At higher levels of autonomy, the platform might be one of many instances overseen by a human controller. At the highest level, the platform might not require any human in the control loop and might be capable of responding to changing conditions reliably automatically and acceptably.

The SEEDPOD model should simulate the various levels of autonomy for each platform and associated manual control links and staffing levels.

5.1.3        Drone Policies and Rules

For all the drones that operate within SEEDPOD, a set of rules will be compiled to which they need to conform. Appendix B lists the 7 basic principles of manned aviation safety that have been developed over a period of many decades. These manned aviation principles are expensive but lead to very high levels of safety. A key question for aviation authorities is how for these need to be re-interpreted and applied for drone and associated un-piloted aircraft.  These will include such things as;

5.1.3.1         Drone regions within which categories of drone are allowed to operate

These regions will have to be established within the model and will be similar to existing classes of three- dimensional manned airspace. These are often referred to as “geo-fencing” in the drone community. However, the fidelity of the regions in SEEDPOD will be much higher and might be much more dynamic and complex. For example, a designated region might only be active at certain times of the day and might and relate to only certain types of platform.

5.1.3.2         Height limits

The most important aspect of drone operations is deciding what sensible operating limits (both upper and lower) make sense in the various drone regions and platform types.

5.1.3.3         Noise limits

A critical aspect of drone operations in an urban or residential area, for example, will be the noise impact on residents/ occupants. SEEDPOD should allow these parameters and limits to be experimented with.

5.1.3.4         Temporal limits (certain times of year of time of day).

For example, bird nesting areas may not allow overflight during the breeding season

5.1.3.5         Traffic density and separation limits

It might be likely that, in the future, there may be a very high demand for drone operations within urban areas. Policy decisions will have to made on the density of operations and separation criteria for drones. There might be technical reasons for a given separation level, for example electronic conspicuity performance limits, or collision avoidance limits based on position uncertainty assurance.

5.1.4        UTM context and rules

SEEDPOD should allow the definition and simulation of a UTM (Unmanned systems Traffic Management) service. A range of operating assumptions will need to be defined ranging from unmanned systems being fully segregated from manned aviation through to fully integrated and non-segregated air-traffic. For context, it will be necessary to simulate manned aircraft operations and interaction with unmanned systems.

5.1.5        Communications and other services

SEEDPOD should be capable of modelling communications availability (satcom, 3G, 4G, 5G, SMS, LoraWAN, WIFI VHF etc) as well as GNSS and other triangulation services. It should be capable of modelling interference and stochastic availability and outages.

5.1.6        Simulation engine

The simulation engine will perform time-step animation and associated analysis of the state of the scenario being studied. Users will be able to define the time horizon, resolution (both spatial and temporal) and features such as weather parameters.

5.1.6.1         Scale

The ambition will be to simulate a large region (the Solent region, for example, including Southampton city) and large numbers of aerial vehicles (possibly thousands).

5.1.6.2         Animation

The model will allow users to study the simulation in a variety of visual representations, viewpoints and perspectives. This will help ensure that the model is validated and checked prior to optimisation runs.

The animation should be capable of being “switched off” for very high speed “batch runs” of simulations.

5.1.6.3         Parallel running and cloud computing

SEEDPOD should be capable of parallel processing so that for very computationally intensive runs multiple cores can be utilised. Ideally SEEDPOD should be capable of exploiting cloud computing facilities and therefore be capable of being scaled in terms of both processors and memory. Users should be able to fully interact with SEEDPOD through web browsers including defining and running models as well as looking at animation and results.      

5.1.6.4         Stochastic capability

The simulation engine will employ random number streams and seeds to allow pseudo random behaviour to be simulated but with fully reproducible replications.

5.1.7        Mission definer

This module will allow a wide range of missions to be defined as well as realistic trigger points for each mission. For example, a mission simulating precision agriculture might be triggered by seasonal dates. Similarly, inshore search and rescue drone operations might have missions driven by weather related criteria. Infrastructure inspection might be related to inspection intervals.

The mission definer will allow users to define platform operating points where drones will land and take-off. Drones will operate within the simulation by obeying the policies and rules defined.

5.1.8        Optimisation

The purpose of SEEDPOD is to be able to objective measure the impact of drones and un-piloted aircraft metrics within a region in terms of the following broad areas;

5.1.8.1         Risk

What is the probability of accidents/ incidents and what is the likely outcome?

5.1.8.2         Economic

What economic effect does the use of drones have in a region?

5.1.8.3         Environmental

What environmental benefits does the service provide overall compared with the existing service?

5.1.8.4         Capacity

What are the capacity limits within a certain scenario?

5.1.8.5         Objective function

SEEDPOD should allow users to define an overall goal in the form of an objective function. This will allow the optimiser to find the best combination of decision variables to best meet the objective.

The objective might be a complex trade-off of all the metrics listed above and may include a number of constraints that have to be satisfied.

5.1.8.6         Results and output

SEEDPOD should allow all of the outputs to be easily shared, analysed, plotted etc using open standards such as XML formatted text files.

Seedpod should provide graphing and data visualisation capabilities.

6         Release strategy

This specification inevitably involves a complex set of requirements in order to deliver a sophisticated and successful tool. It is therefore a requirement that a release strategy is planned to allow early releases of versions of the tool that have basic functionality for user testing and assessment. The tool should be designed so that extra functionality should be capable of being progressively added seamlessly. Releases should be made early and certainly at least every 6 months during the development life cycle.

Software should be developed professionally and using good software development practices with well partitioned and modular code.

7         Documentation

Code should be fully and professionally documented. User guides and manuals should be developed and delivered as part of the software development process.

8         Open Source

SEEDPOD will be released as open source code.

Appendix A

Existing work

NATS FloSys (https://www.nats.aero/services/airspace/flosys-flight-optimisation-system/ )

Dynamic airspace optimisation (Standfuß; DLR https://link.springer.com/article/10.1007/s13272-018-0310-7 )

Agent based modelling of UAVs (Schuman; https://eprints.soton.ac.uk/202055/)

Optimization of Airspace & Procedures in the Metroplex (OAPM) (FAA; https://www.faa.gov/air_traffic/flight_info/aeronav/procedures/oapm/ )

Air route network optimization in fragmented airspace based on cellular automata (Wang; https://www.sciencedirect.com/science/article/pii/S1000936117300900 )

EPSRC DECODE project; Value optimisation of UAV systems (Gorissen; https://arc.aiaa.org/doi/abs/10.2514/1.C032153 )

Optimised Airspace User Operations – SESAR 2020 project PJ07 (OAUO) (https://www.eurocontrol.int/articles/optimised-airspace-user-operations-sesar-2020-pj07-oauo )

NASA’s CERTAIN to use NextNav 3D geolocation tech for urban drone operations (https://airtrafficmanagement.keypublishing.com/2018/11/13/nasas-certain-to-use-nextnav-3d-geolocation-tech-for-urban-drone-operations/ )

UAS ATM integration operational concept – Eurocontrol (https://www.eurocontrol.int/sites/default/files/publication/files/uas-atm-integration-operational-concept-v1.0-release%2020181128.pdf )

Concepts of Airspace Structures and System Analysis for UAS Traffic flows for Urban Areas (Jang; https://utm.arc.nasa.gov/docs/2017-Jang_SciTech_2017-0449.pdf )

European ATM Master Plan; Roadmap for the safe integration of drones into all classes of airspace (https://www.sesarju.eu/sites/default/files/documents/reports/European%20ATM%20Master%20Plan%20Drone%20roadmap.pdf )

A Novel Air Traffic Management Decision Support System Gardi PhD thesis (https://researchbank.rmit.edu.au/eserv/rmit:161950/Gardi.pdf )

Airport and airspace simulation modelling http://www.simscript.com/solutions/transportation/Airport_and_Airspace.html

NASA UTM Research TCL (Technical Capability Levels)(Kopardekar, 2016)

Position Uncertainty Volume (PUV; Ramasamy)

http://www.atmseminarus.org – for previous ATM and UAS research – Next conference in Europe this summer.

http://2018.dasconline.org/ – Digital Avionics Systems Conference DASC – last conference in London last year.

SESAR Innovation D

Singapore urban delivery  – Airbus, and Singapore ATMRI (Nnangyang University, Professor Kim Low Hutt)

Onera – French transport research consultancy – did a capacity study of urban airspace for drones how many could you fit in to a volume of airspace.

Norkopping Uni in Sweden jonas.lundberg@liu.se has led a lot of the research for LFV – the Swedish NATS equivalent.

Billy Josefsson, LFV Research & Innovation, along with Danwei Tran Lucian, PhD student – Linköping University, spoke about the future air traffic management of drones.

LFV operates two projects together with Linköping University; UTM 50 and UTM OK. UTM 50 visualizes future drone traffic and how air traffic management as regulation and services can create a monitored and safe airspace ahead. UTM OK analyzes airspace issues. The project is the basis of an automated route planning system for the allocation and monitoring of airspace utilization.

“LFV operates for both route planning, safety and concept development, also through integrating knowledge and results with our delivery of air traffic management services, we can continue to be a leading operator,” said Billy Josefsson.”

UTM Implementation In Cities – Sampled Side-Effects

https://www.researchgate.net/publication/329563948_Urban_Air_Traffic_Management_UTM_Implementation_In_Cities_-_Sampled_Side-Effects

Jonas Lundberg dept. of Science and Technology Linköping University Norrköping, Sweden

Billy Josefsson LFV Air Navigation Services of Sweden Norrköping, Sweden billy.josefsson@lfv.se

https://www.futuristspeaker.com/business-trends/37-critical-problems-that-need-to-be-solved-for-drone-delivery-to-become-viable/  …….

1. Designated Delivery Spots – Much like mail delivery, drones will need designated places for package delivery. Commercial delivery to businesses will have different guidelines than home delivery.

2. Durability – Manufacturing drones durable enough to make 100 deliveries between scheduled maintenance and 10,000 flights over their lifetime will be an absolute necessity.

3. Conditional Awareness – Drones will invariable fly into unusual situations, and whether it’s swarms of bees, bird attacks, lightening strikes, or signal jammers, they will need to alert operators of problems as soon as they arise.

4. Black Boxes – Much like today’s commercial aircrafts, whenever a drone crashes, some sort of signaling device will be needed to allow for follow-up investigation and cleanup.

5. Maintenance Plans – Today’s hobbyist drones seem like simple contraptions, but higher end delivery drones will need a consistent schedule for prop replacement, motor alignment, sensor checks, controller board cleaning, etc.

6. Override Kill Switch – Wireless signals are far from perfect. If a signal is lost, hacked, or hijacked, the drone must either return home or be removed from danger.

7. Drone Classification System – Drones are being created in thousands of different shapes and sizes with thousands of different capabilities. A comprehensive classification system will be needed to properly manage and regulate this industry.

8. Cargo Classification System – Cargo classification systems applied to ground-based shipping will need to be revised for the more volatile conditions associated with remote controlled airborne vehicles.

9. Drone Insurance – Drones, drone cargo, and drone businesses will soon become the largest new market for insurance companies.

10. Vehicle Licensing – Every drone that falls within certain classification guidelines will need to be licensed and insured.

11. Pilot Licensing – Those who fly drones will need to be tested and licensed in a less rigorous but similar way that airplane pilots are tested today.

12. Operator Licensing – People who load and unload cargo onto flying drones will also need to be licensed.

13. Weather Contingency Plans – Every drone will have to deal with extreme weather at one time or another. Any condition ranging from wind, to rain, snow, hail, extreme heat or extreme cold, will need a contingency plan for both the retrieval and safe delivery of the cargo.

14. Privacy Rules – Privacy means different things to different people, but flying drones with cameras, scanners, and sensors give nefarious people far more capabilities than ever before. Privacy rules will need to be established sooner than later.

15. Security Rules – Once a famous person’s delivery address becomes known, they run the risk of receiving unwanted packages, solicitations, threats, and even things like chemical attacks.

16. Drone Spam Rules – Much like junk mail and spam email, flying drones open up the possibility of receiving everything from annoying products samples to mean-spirited pranks. Rules for “drone hate crimes” and “drone bullying” will soon follow.

17. Noise Guidelines – The larger the drone and the greater the distance it has to cover, the larger the engine it will need to operate. Since electric drones only cover short distances, some form of petrochemical fuel will be needed, and these vehicles will be noisy. Rather than waiting for 10,000 communities to imposed their own one-off noise ordinances, it would be better for the industry to be proactive in this area.

18. Automated Here-to-There Delivery – Drone delivery only becomes a mass-market affordable option when human operators are removed from the equation.

19. Grasp and Release Mechanisms – People who set a package out front, wanting to send it across town, will require a pickup drone capable of automated grasp and release.

20. Aerodynamic Packaging – Packages attached to the bottom of a drone will need to be far more aerodynamic than the rectangular boxes most commonly delivered today.

21. Fly-Drive Capabilities – Because of trees, porticos, awnings, and overhangs, drones may need the ability to land on open space and drive to the appropriate delivery spot.

22. Collision Avoidance Systems – With the potential of flying into everything from power lines, to trees, windmills, Christmas decorations, and other UAVs, a comprehensive collision avoidance system will be necessary.

23. Crowded Skies Navigation System – At some point in the future there may be as many as 10,000 drones flying over a city in a given day. Not only will they need to avoid flying into buildings, trees, and commercial aircraft, they will need to avoid other drones as well.

24. Drone Operating System – An operating system is the most important software that runs on a computer because it defines how it functions. Computer buyers typically will choose between Android, iOS, Linux, or Windows for their operating system. Since drones have a different role and purpose, they will require an entirely different kind of operating system.

25. Shot from the Sky Recourse – Many disturbed individuals will view drones as a “form of target practice.” Drone owners and operators will need recourse for these situations.

26. Political Awareness – Paranoia is already rampant when it comes to all the bad things people can do with drones. For this reason its imperative that politicians be given special attention so they can understand the cost-benefit ratio associated with any of their decisions.

27. Consumer Awareness – Rather than letting the news media define the industry, this emerging industry needs to be proactive in defining itself.

28. Education for the Drone Police – Police will not only employ drones to assist in managing public safety, they will also use drones to monitor other drones. Drones are far more versatile and faster to deploy than virtually all other options officers have at their disposal.

29. Education for Drone Lobbyists – Drones will become one of the most highly regulated industries of all times. It is not too soon to start educating the influencers.

30. Education & Certification for Drone Pilots – With all their different configurations, styles, and function, drone pilots will require far different training than airline pilots do. Currently there are very few simulation programs available for practice.

31. Education for Drone Maintenance and Repair – People who service and fix drones will be in hot demand in the near future.

32. Drone Financing – As the need for instrumentation and safety equipment mushrooms, delivery drones will become far more expensive. As a result, drone financing will become a hot new area of business in the near future.

33. Flying Drone Bill of Rights – Do people have the right to “keep and bear drones?”

34. Docking Systems – People will eventually not want packages delivered onto their driveways. For example, any pizza left on a driveway becomes an open invitation for cats, dogs, and other stray animals. A better option would be to have some sort of docking system that would allow the drone to land and deliver the package into a secure area.

35. Better Battery Tech – Battery technology has not progressed nearly fast enough for the drone industry. Even gas-powered drones will likely need batteries to fly through “quiet zones” such as hospitals, nursing homes, and environmentally sensitive areas.

36. Airbag Crash Protectors – Accidents will happen and on occasion, drones will indeed fall out of the sky. To prevent large drones with heavy or dangerous payload from causing serious damage to people and property on the ground, some form of rapidly inflating airbag will be needed.

37. Invisible Fences – There will be many no-fly zones around the world and these zones will need the equivalent of an “invisible fence” to keep intruders out.

Appendix B