Notice for Proposed Rule Making for CASR PART 101

1
Remotely Piloted Aircraft Systems – Changes to Rules
CASR Part 101
Respond to CASA by 16 June 2014

CASA has made notification for Proposed Rule Making for CASR PART 101

Advisory Circular 101, remotely piloted aircraft systems rules will affect the application and utilization of Unmanned Aircraft Systems.

CoaX Helicopters believes that safety is the highest priority for RPA operations.

In so saying, CoaX also believes that a number of statements contained in the Notice require further examination prior to promulgation.

A detailed discussion is provided below

Please provide feedback directly to CASA.
Respond to CASA by 16 June 2014.

The reference web page is
http://casa.gov.au/scripts/nc.dll?WCMS:STANDARD::pc=PC_102028

The content of the revised rules are shown here
http://casa.gov.au/wcmswr/_assets/main/newrules/ops/nprm/nprm1309os_annexc.pdf

Click on the CASA icon below.
CASA-UAS-Logo_2

VLOS / EVLOS Discussion
Hypothesis

That the CASA’s policies requiring that the aviation safety regulations for VLOS, EVLOS and Collision Avoidance are satisfied.

REMOTELY PILOTED AIRCRAFT SYSTEMS NPRM 1309OS
dis_IMG_0328
Foreword
This Notice of Proposed Rule Making (NPRM) is issued by CASA with a view to ensuring that
Australian aviation safety requirements are current and appropriately address safety risks.
CASA’s policies require that the aviation safety regulations must:
• be necessary to address known or likely safety risks
• provide for the most efficient allocation of Industry and CASA resources
• be clear and concise
• where appropriate, be aligned with international standards and drafted in outcome based terms

When tested as described in the discussion below, the hypothesis fails on two counts
1. It does not address ‘known or likely safety risks’ and
2. It is not ‘clear and concise’.

Assumptions
In order to ensure adequate understanding is provided to all stakeholders as described in AC 101 Remotely piloted aircraft systems, it is necessary to make a number of assumptions.
1. RPA operators may have no more than a few hours of aeronautical experience.
2. RPA operators may be operating in or near airspace used by manned aircraft.
3. RPA will require unambiguous, self-contained (ie described within the wording of AC 101) and self-explanatory definitions and methodologies written into AC 101.

Definitions
Several additional definitions are required to describe RPA operations that are currently inadequate in the draft AC 101 Remotely piloted aircraft systems
The following definitions and descriptions are provided:
1. Collision Path.
Definition: A course or path between two or more objects which, if continued at the current rate, will result in a collision.
2. Determination of a Collision Path
Definition: A collision path results if a constant bearing is held between two or more moving objects and those objects are closing toward each other.
Note – the only CASA reference to collision avoidance and methodology to detect a collision is contained in the following document.

CAAP 166 2(1) – Pilots’ responsibility for collision avoidance in the vicinity of non-controlled aerodromes using ‘see-and-avoid’

7.4 Pilots should also be aware that two aircraft converging on a point have the potential to remain fixed in one or both pilots’ field of view, i.e. their relative position (in the windscreen) won’t change until moments before impact.

3. Collision Avoidance
Definition: Manoeuvring an object, specifically an aircraft, to ensure a constant bearing is not maintained or, does not subsequently occur, between the aircraft being controlled and a conflicting object(s) when there is closure between those objects.
4. Collision Avoidance Responsibilities
Definition: Operations of an RPA to ensure Collision Avoidance
5. Separation
Definition: Separation is the concept of ensuring acft maintain a prescribed minimum from another acft or object, whilst meeting the associated conditon(s), and requirements of the standard. This is currently specified in MATS (Manual of Air Traffic Service)
This definition may need more collaboration and discussion to describe more fully a suitable definition for Separation utilizing RPA operations.

Proposal
Extended Visual Line of Site (EVLOS)
An operation where the Remote Pilot (RP) does not have direct visual sight
with the RPA. However, with the assistance from trained RPA observers, the
RP is still able to ensure safe operation of the RPA by avoiding collisions with
other traffic. At all times, at least one of the RPA observers is to have direct
visual sight of the RPA and is to be able to communicate with the RP in order
to manage the flight of the RPA and for the RP to meet his/her collision avoidance responsibilities

Analysis
There are three methods to determine that an aircraft is on a collision course
1. Visually, by observing from a moving platform that another object is on a collision course. This is also referred to as a pure pursuit course. When making this determination the observer will note that there is no crossing motion with respect to the platform’s frame of reference at the time of observation.
2. By utilising an ACAS system. This will require all conflicting aircraft to electronically communicate with each other and use algorithms to calculate that a collision will occur. The communication will require knowledge of altitude and vertical and horizontal velocity.
3. By utilising on-board detect and avoid system(s). This may be systems such as LIDAR, pixilation discrimination, sonic detection and tracking. All of these systems however require that the detection devices are mounted on the platform. The subsequent processing will ultimately calculate that a constant bearing exists and the platform is on a collision course.
An approximation can be made by a visual observer when not co-located with the platform under RPA control. This will require consideration of the following
1. The size of the RPA and any intruder aircraft ( the observer will need aircraft identification and recognition skills)
2. The speed of the RPA and any intruder aircraft
3. The angular distance between the RPA and any intruder aircraft
4. The rate of change of the mils subtended by the RPA and any intruder aircraft
5. Whether the aircraft, particularly regarding RPA to RPA deconfliction, are flying in a direction consistent with the natural fore-aft axis of the aircraft
6. The altitude of all aircraft
7. The range of the aircraft (particularly difficult with a featureless or indistinguishable background)
8. Whether the intruder is turning, climbing or descending
9. Whether the RPA is turning, climbing or descending
10. A determination that ‘Return to Base’ or ‘Land Immediately’ are not actions that will increase the actual risk of collision
The outcome of this approximation technique will at best only be a good guess. Unless the sensing or detection is done coincident with the RPA platform, an accurate determination of a collision path existing cannot be made.

The requirements of VLOS and EVLOS whereby the RP is to ‘ensure safe operation of the RPA by avoiding collisions with other traffic’ cannot be achieved. There is no capability for either an RP or an observer, trained or otherwise to determine that a collision course exists.

As the requirements of collision avoidance cannot be assured unless collocating sensing or detecting, the only other way of deconflicting is by using separation techniques.

Separation can be Strategic or Tactical.

Strategic separation can be reasonably achieved by airspace usage notification. Utilization of dedicated protected airspace will provide some degree of confidence in separation. This can be extended by use of NOTAM. For all intents and purposes this is segregation of manned and unmanned operations.

Tactical separation can be achieved in flight when it becomes apparent that one or more intruders exist. By recovering to the departure point, landing immediately or flying away from an area of confliction, separation should occur.

There are two problems when employing Tactical Separation:
1. Speed differential. It may not be physically possible fly the RPA away from the intruder in sufficient time.
2. Unpredictability of the intruder. If the RPA cannot determine accurately intruder changes in direction the RPA may inadvertently fly into the intruder.

RPA operators may well have been successful avoiding collisions in the past but that does not mean that it is doing collision avoidance. Under VLOS/EVLOS separation is the only way to prevent a collision. Avoiding action may result from a formal ATC process, a detect and avoid process, a sense and avoid process or a segregated airspace process. Without collocated constant bearing determination, the PR operator cannot assume that a manoeuvre will resolve a collision potential. Simply making a decision to ‘land now’ or ‘return to base’ does not provide any mitigation of collision but it can be mitigated by way of separation. Part of the problem stems from the fact that we often consider an intruder to be a fixed wing manned aircraft. If the intruder is another RPA, non-directional in aspect, significantly different in size and simply pops up out of the blue the RP’s risk mitigation might say ‘land now’ or possibly ‘descend now’. It can be easily demonstrated that the RP or observer could not differentiate between the intruder aircraft and the RPA as to whether the intruder was closer or further away, whether it was above or below or whether it was flying on a convergent or divergent path and further whether landing will increase or decrease the chance of collision.
Tactical Separation is easy to describe and define and much better than a Big Sky approach. It provides a greater degree of confidence in avoiding a collision and is achievable unlike the current collision avoidance philosophy.

EVLOS
‘RPA observers is to have direct visual sight of the RPA’
The requirement for an RPA Observer to maintain direct visual line of sight provides a false expectation that this will support collision avoidance.
I have provided a scenario described in Annex 1 that highlights the inappropriateness of having eyes on the RPA. In this example an EVLOS observer is positioned to maintain eyes on the RPA. The scenario demonstrates that the observer should in fact be positioned over 2km away from the RPA to provide sufficient flight time to achieve separation.
As has already been discussed, it is not possible to ensure collision avoidance when using non- collocated RP’s or observers. The best that can be done is to plan some type of Tactical separation when the observer detects an intruder.
A calculation must be made to determine the longest time to land or time to achieve Tactical separation for the RPA. Knowledge of likely intruders should also be attained. Observers should be positioned so that should an intruder be identified, there is sufficient time available for the RPA operator to complete the Tactical separation. The observer’s task should be to scan the area for likely intruders subsequently providing adequate notification to the RP that an intruder is approaching. Keeping watch on the RPA will not decrease collision potential.

Conclusion
Requiring an RP to provide Collision Avoidance when operating using non-collocated visual line-of-sight procedures is unachievable.
Understanding the physics of collision avoidance is not well understood.
EVLOS and VLOS provisions as written in the draft AC 101 will persist as a known safety risk.
With a significant number of RPA operators who may have little aeronautical experience it is imperative that definitions are comprehensive, clear and unambiguous. Draft AC 101is deficient with several omissions and a number of undefined requirements.

Recommendations
That the terms “Collision Avoidance” and “Collision Avoidance Responsibilities” in the context of non-collocated visual line-of-sight procedures should be removed
That EVLOS be completely rewritten. The policy should put emphasis on Tactical Separation to minimise collision risk. Observers should be trained to scan for intruders, not watch the RPA.
That AC 101 definitions must be expanded to include
1. Collision Path
2. Determination of a Collision Path
3. Collision Avoidance
4. Collision Avoidance Responsibilities
5. Separation
After incorporation of the recommendations, the hypothesis must be retested to ensure that CASA’s policies required by the aviation safety regulations for VLOS, EVLOS are satisfied

Example Text
Visual line-of-sight operation
An operation in which the remote crew can maintain direct visual contact with the aircraft, aided only by spectacles or contact lenses (not binoculars or telescopes etc.) to manage its flight and meet separation expectations using Strategic and Tactical Separation techniques.

3.1 Standard RPA operating conditions
Visual Line of Sight (VLOS), i.e., an operation in which the remote crew aided only by spectacles or contact lenses (not binoculars or telescopes etc.) can maintain direct visual contact with the aircraft, to manage its flight and meet separation expectations using Strategic and Tactical Separation techniques.

Extended Visual Line of Site (EVLOS)
An operation where the Remote Pilot (RP) does not have direct visual sight with the RPA. However, with the assistance from trained RPA observers, the RP is still able to ensure safe operation of the RPA and meet separation expectations with other traffic using Strategic and Tactical Separation techniques. At all times, at least one of the RPA observers is to scan the airspace surrounding area of the RPA operation at a range that will allow effective Tactical Separation and is to be able to communicate with the RP in order to manage the flight of the RPA and for the RP to meet his/her meet separation expectations using Strategic and Tactical Separation techniques.

4.2 Extended Visual Line of Site
Extended Visual Line of Site (EVLOS) is where the RP does not have direct visual sight with the RPA. However, with the assistance from trained RPA observers, the RP is still able to ensure safe operation of the RPA and meet separation expectations with other traffic using Strategic and Tactical Separation techniques. At all times, at least one of the RPA observers is to scan the airspace surrounding area of the RPA operation at a range that will allow effective Tactical Separation and is to be able to communicate with the RP in order to manage the flight of the RPA and for the RP to meet his/her meet separation expectations using Strategic and Tactical Separation techniques

dis_EVLOS_4
Annex 1 – EVLOS Scenario
Annex 1

A scenario that highlights deficiencies of the current EVLOS proposals .

Let us assume we have a photographic mission to examine Long Shore Current and Sediment Drift. The task requires 3 grid patterns to be flown 1, 2 and 3, north to south, in the region of the Skillion near Terrigal.

I will make a few assumptions and place some constraints to fit the scenario. It will be possible to raise counterpoints but that is not the purpose. I am very confident other similar locations and conditions could be found to build a similar context.

Assumptions
1. RPA TAS = 50kts (Note that for slower RPA’s the equation becomes worse)
2. 20kt wind NNE
3. Cloud generally clear with scuddy low level build ups off the coast.
4. Height of launch area 150’ AMSL
5. No other suitable recovery area. Any landing other than in the recovery area will result in damage or destruction of the RPA

Task
1. To fly a 400m * 400m grid pattern at 150’ AMSL over a 2 hr period and collect photographic data of the overflown area.

Grid 1 has been completed and Grid 2 is now being flown.

H is the launch site and X2 and X3 are the locations of the EVLOS observers.

The risk assessment considers the possibility that a light aircraft may transit the survey area. The restriction on light aircraft operating in the area would be an expectation that the aircraft did not fly below 500’ AMSL. As the RPA schedule calls for flight at 150’ AMSL segregation is reasonably assured.

The Grid 2 observer is positioned at X2 to provide ‘eyes on RPA’ whilst the RPA is operating in Grid 2.

Part way through the mission when the RPA is at the most southern point of Grid 2, a light aircraft is noticed flying southbound along the coastline.

The mission risk mitigation says that a ‘Land Immediately’ protocol is to be applied as soon as any intruder is identified.

The RP initiates immediate return to home. The profile has been defined as
1. Climb immediately to 50m above the launch altitude
2. Track directly to the landing point
3. Use maximum speed
4. Land at the launch site as soon as possible

The pilot of the intruder aircraft has up until now been in reasonably clear weather. He now notices the weather is deteriorating in the region east of The Skillion. He makes a decision to remain visual and stay as close to the coast as possible. He descends to 300’ and tracks along the red dashed line to remain visual and clear of cloud.

With the 20 kt wind the RPA will have a ground speed of about 30kts. (15 m/s)
It will take about 30 seconds for the RPA to return to the landing site.
The aircraft is flying at 130kts, ground speed about 150kts. (77m/s)

In simple terms the intruder is flying 5 times faster than the RPA. If you plot this out the intruder would need to be identified further north than the photo image actually depicts (a little over 2300m). In other words you would need an observer near Terrigal Lagoon at position XE to adequately detect the intruder with sufficient time to land the RPA.

As it stands, the intruder will fly directly through the flight path of the RPA at the same altitude.

Humans do not have the physical capacity to evaluate a collision risk if they are not collocated with one of the aircraft. Quite possibly the intruder may track directly overhead the RP. The RP will not be able to determine if the intruder is higher or lower than the RPA or whether the random manoeuvring of the intruder will mitigate or enhance the risk of collision should the RPA attempt evasive manoeuvring.

Note also that for every EVLOS observer that extends the range of your VLOS, you increase the time of flight for recovery. This will increase the volume of space an intruder can penetrate and therefore creates an increased risk. The second observer is depicted in Grid 3 at the position X3.

For your consideration.

Figures

48 knots (55 mph; 89 km/h)
20 kt hw
Approx 30 kt g/s = about 15 mps
400 m / 15 mps = just under 30 secs
AC speed 130 kts + 20 kts wind = 150 kts = 77 m/s
Minimum detection range = 2310 m