Avetics Safety Series - Understanding Multicopter Redundancy

We are introducing a safety series, in-depth observations from our engineering teams to share drone safety insights from engineering perspectives. These insights are based on our operational and R&D experience.

In this article, we will be discussing multicopter redundancy. When it comes to safety, one key consideration is propulsion redundancy. However, not all drones are made equal and it is important for evaluators to understand the key differences between drone motor configurations and the specific behavior of each configuration during motor failure.

Propulsion basics

Propeller is attached to a motor which spins either anti-clockwise or clockwise. This creates thrust - but also a clockwise or anti clockwise moment.

Moment - In simple terms, is a measure of its tendency to cause a body to rotate about a specific point or axis.

The moment created will be in the opposite direction of the propeller spin direction. This moment if not cancelled, will cause the drone to spin. If the thrust is too much on one side of the drone it will cause the drone to flip to the other. Hence thrust must be balanced on all sides for stability.

Quadcopter

By far the most popular configuration. Quadcopters have 4 motors usually arranged in the X configuration. Quadcopters might still be controllable when 1 side of a propeller breaks but when 1 motor fails, the quadcopter will flip and crash.

Thrust

When M2 fails, there is no thrust from M2. The drone falls towards M2, just like what will happen to a 4 legged square table with one broken leg.

 

Hexacopter

This is also a very popular design and is usually used when the drone needs to lift a heavier load.

Thrust

If M2 fails, there is no thrust from M2. To maintain level flight, M1 and M3 spins faster to make up for the lost thrust from M2. The drone is able to maintain level flight.

 

 

Moments

M2 spins clockwise, which generates anti-clockwise moments. When M2 fails, there is a loss of anti-clockwise moment, causing a net clockwise moment.

To make matters worse, while M1 and M3 spin faster to compensate for the lost thrust, they create more clockwise moments.

The platform has a net clockwise moment and will start spinning in a clockwise direction.

In modern drones with Intelligent Course Lock feature, the drone will still be controllable by a moderately skilled pilot since the computer onboard will take pilot input command (for example to pitch back) and transform it into actual commands relative to the drone heading (which is ever changing since the drone is spinning).

Note that this works only when there is solid GPS and compass signal as those algorithms need GPS and compass data to determine position. Without this feature, only an expert pilot will be able to land the drone since he will have to constantly change inputs relative to the drone heading just to command the drone in the desired direction.

 

Octocopter

The most expensive design will be the Octocopter.  Howeverm, compared to the quadcopter. This configuration is less efficient in propulsion, this is because for an equivalent size a Quadcopter will have more space between motors to spin longer propellers.

Thrust

If M2 fails, there is no thrust from M2. To main level flight, M1 and M3 spins faster to make up for the lost thrust from M2. The drone is able to maintain level flight.

 

Moments

M2 spins clockwise, which generates anti-clockwise moments. When M2 fails, there is a loss of anti-clockwise moment, causing a net clockwise moment.

While M1 and M3 spin faster to compensate for the lost thrust, they create more clockwise moments.

The platform has a net clockwise moment and will start spinning in a clockwise direction.

To prevent this, M4 and M8 spins faster to create additional anti-clockwise moments. When the additional anti-clockwise moments cancels out the net clockwise moment of the drone, there is no more net moments, and the drone does not spin.

M4 and M8 spinning faster will not cause the drone to pitch or roll, and still have level flight as they are opposite each other.

Thrust

M4 and M8 spinning faster increases the total thrust of the drone. To prevent the drone from climbing, all motors will spin slightly slower to decrease thrust in order to maintain height

The effect is the drone maintains thrust and heading. A trained pilot will be able to land the drone without any advance algorithms or strong GPS signal.

Conclusion

The Octocopter design presents the best chance of recovery when 1 motor fails. While the Hexacopter depends on how skilled the pilot is and whether advanced flight algorithms with strong GPS signals are available. The Quadcopter will most certainly crash.

Drones carrying people - death awaits

The Trend

There seems to be more and more projects involving drones carrying people. Ehang 184, a X8 design multicopter announced at CES 2016, is dubbed the air taxi of our time and is launching flights in Dubai soon.

Source: Popular Mechanics

Volocopter, a German based company is also launching their version as well with 18 propellers. There's also the Kitty Hawk, a project backed by Google's cofounder Larry Page, that looks more of a cross between a motorcycle and a drone. 

Source: Engadget

The Technology

Almost all of them share similar base operating concept as a small drone. A central processor with orientation and speed sensors giving commands to vary speeds of various motors. The difference is that they use substantially larger motors and probably better quality flight electronics. Unlike a plane or helicopter, the movement of these large drones are not controlled by mechanical means of varying some sort of control surface. Instead they rely on changing rotating speed of the motors. Therefore control is almost always electrical. This is important because unlike an aircraft with mechanical control as backup, these drones have to rely on electrical means which could be the cause of failure in the first place. Essentially redundancies highly dependent on the electronics not screwing up. Note that Volocopter and Ehang does not provide any control means for the passenger, it is essentially fully autonomous. A helicopter can still be controlled even if the propulsion fails by means of autorotation but not these drones.

Source: Transport Canada

Regulations

Unlike the commercial aviation industry, standards are not yet well established in this sector. The tests required are not well defined and robustness of the system is not proven.  It will probably take a few deaths before the industry becomes more regulated. This is not a morbid prediction but base on history of the aviation industry when it first started in the 1900s. Popular Mechanics has an excellent article that summarises changes made to aviation after crashes.  This will probably be the same case of drones carrying people.  Still not convinced? Ehang actually makes consumer drones as well, see here for a collection of Ehang Ghost drone crashing. 

It takes two to fly : Part 1

Many of our operations with our larger custom-made drones comprises of a two man crew, the pilot and Gimbal operator. To the uninitiated, the pair cuts a strange sight, one looking up at the sky and the other seemingly looking at the ground - actually looking at the monitor of his or her screen, observing the camera feed that is being beamed back from the drone in the air.

Typical dual operator drone.

Basically, the pilot controls the drone, while the Gimbal operator controls the independent motorized camera suspending from the bottom of the drone - which can swivel left to right and tilt up and down.

Teamwork between Gimbal operator and pilot is an undervalued aspect of drone operations, and is a mix of a healthy dose of mutual respect, trust, constant communication, understanding of each others limitations, a communication language and a drop of chemistry. 

Sounds like a relationship? Well it really is, and more often then not It is what separates a competent team from a great team, and likewise, acceptable results from great results.

On this topic, there are many things to discuss and share, but some points may have jumped out at you, such as:

Understanding limitations:

Drone operations, are high stress operations, each operator has multiple roles.

Pilot :

1. flying drone
2. monitoring weather conditions
3. monitoring drone telemetry
4. scanning of potential obstacles and hazards
5. listening to Gimbal operator instructions and translating them to drone movements

Gimbal operator :

1. liaising with client, communicating with pilots for drone movement
2. controlling camera movement
3. assisting pilot with telemetry read out
4. controlling the ground around take off and landing points
5. managing the public passersbys (managing curious onlookers is an almost full time job)

All these are aimed at getting the drone maneuvered to a right point to ensure that objectives are achieved, be it the best angle for an architectural shot, or the right angle to look through a nook and cranny to inspect bolt fittings on a tower crane - all while operating in a limited 15+ minute window before the battery runs out.

Limited batteries can be brought to the field, which piles on more stress, exacerbated by a client breathing over your shoulder(s). Each flight has to count.

In general, these the two staff sometimes have conflicting objectives, the pilot does wants to get the bird down to the ground safely, the Gimbal operator wants to stay in the air to grab more photographs/videos/scans.

These conflicts can naturally results in disagreements. Many a story has circulated within the industry of pilot and Gimbal operator squabbling (in front of a client!) a more drastic one I heard were teammates walking away from each other in the middle of a job.

This is where respect for limitations comes in.

From my perspective as a Gimbal operator, I usually am the one who has a larger risk tolerance, to push my pilot to try complex movements to get more dynamic shots, whilst, the pilots are naturally more conservative, they want to make sure safety is the top priority. It is a good check and balance and a healthy dynamic.

Usually, this is managed by a pre-flight discussion between partners before take off, to talk through the process with each other. Considerations should be discussed before the drone is in the air and precious battery amps are wasted.

But more often then not, the perspective does change in the air, the drone angle of attack might need to be altered from plan. Quick discussion and some bargaining between the two might happen. "Perhaps we push 5 meters more?" "Lets push the battery 0.1 more volts before coming back"

However, irregardless of seniority, the pilots have the final say. If a firm "No" is elicited, all discussion is over, and I will communicate to the client what issues are faced, and safety cannot be compromised.

Another ironic thing is that when the pilot is flying, the Gimbal operator is the one with more information, as he has control on the camera and can view direct surroundings of the drone more precisely, observing potential obstacles, such as trees, buildings, etc.

Pilots can get nervous when the drone starts to merge into the background, as drones are relatively small, past a certain point it is difficult to tell its distance relative to its surroundings. Parallax error ensues.

This is what the pilot sometimes sees, Is the drone in front or behind building? Above or below crane? How sure are you?

But the Gimbal operator knows. He can use the camera feed he is receiving to judge distance more accurately. Most of the time, there is usually still a large distance, of 30 meters. But it is just hard for the pilot to tell with the naked eye.

The Gimbal operator has to reassure the pilot of the sufficient distance for maneuvers, and the pilot in turn has to trust that the operator is keeping his camera on a swivel, being the pilot's eyes during these types of situations.

So this is one aspect of a healthy and effective drone duo at work, I will share more in upcoming posts.