It’s the scenario no pilot wants to imagine: A medical emergency that renders him or her incapacitated, leaving passengers helpless, panicked and facing an unthinkable outcome. But Garmin has changed the calculus on such an event. With its revolutionary Autoland system, the aircraft will navigate to the closest, suitable airport, select the most appropriate runway in regard to winds and weather, and safely land and stop the aircraft on the runway. This happens with a simple push of a prominent red button on the flight deck.
Publicly announced in late October, Autoland is scheduled to first be certified on the Piper M600SLS, Piper’s flagship single-engine turboprop, by the end of the year. It is also scheduled to be certified on the Cirrus SF50 Vision Jet by yearend.
In August, Garmin invited me to test out the Autoland feature at its flight test headquarters at New Century Air Center (KIXD) onboard its M600 testbed. After a thorough brief and review of the system, we climbed aboard the M600 for a demo flight. With flight test pilot Erik Sargent in the left seat, I settled into the right seat, putting myself in the position of a passenger.
After departing Runway 18, we climbed northwest of the airport to an altitude of 4,000 feet. With the aircraft on autopilot, Sargent invited me to lift the clear plastic guard and activate the Autoland button. Immediately the familiar G3000 displays were replaced with clear communications that “Autoland is Active.” At the same time, the airplane had activated the GPS LNAV approach for Runway 18 and made a 180-degree toward KEZNU, the final approach fix. The autothrottles slowed the aircraft as it descended, and as the aircraft turned onto the final approach course, the gear and approach flaps extended and the aircraft settled into a stabilized, 110-kt descent to the runway. Meanwhile, the displays and aural instructions provided continual updates and guidance on ensuring seatbelts are buckled and loose items are stowed. On the final approach segment, the MFD displayed the message, “Approaching Destination Airport. Prepare to Land.”
This is where things get interesting, especially for a pilot accustomed to disconnecting the autopilot at decision height or on short final. Sargent assured me that it wouldn’t be necessary today. As we crossed the threshold, the autothrottles retarded power and the plane gently began to pitch up for the flare. Touching down firmly on the mains, the aircraft tracked perfectly down the runway as it brought the nose down and braked aggressively. Once at a complete stop, Autoland normally would initiate engine shutdown, but on this demo, it was prevented from performing that function.
Although it landed slightly left the painted centerline, just as other aviation journalists have reported, the airplane was dead center on the synthetic vision GPS waypoint. (The paint stripes are not perfectly aligned with the GPS runway center point).
Throughout the demo, the automated announcements provided reassuring updates of the aircraft’s progress toward its destination and even provided helpful pre- and post-landing instructions.
“Well, what do you think?” Sargent asked.
I could only conjure up one word: “Amazing.”
Deep-Dive into Development Background
Garmin first began imagining an Autoland system in the early 2000s and began working on it in earnest in 2010 using a Columbia single-engine aircraft as its test platform. The goal at first was to simply get the aircraft on the ground to save the lives of those onboard. As development continued, the goal changed to not only save lives but to preserve the integrity of the aircraft itself.
Development was divided between three main groups within the company: the flight controls group, which was responsible for autothrottle, auto-braking, auto-steering and other mechanical automation aspects; the display software team, which developed the algorithms for the system’s decision-making, routing and the most difficult part: the landing; and finally, the certification arm of Garmin worked in concert with Piper and Cirrus to make the entire system on the M600 and Vision Jet a certified reality.
On the Columbia, Garmin further refined its 3-axis autopilot, then delved into developing an autothrottle by developing a servo that could manipulate the throttle to control speed.
“From an autopilot perspective, we spent a lot of time on the basics of the 200 feet down to the runway surface: the flare, the low-speed ground effect, the landing. Then we tackled the last piece, which was the ground control, the aircraft’s braking and steering,” said Ben Patel, the team leader for Garmin’s flight controls group.
To test and perfect the landing sequence, engineers modified the aircraft’s airport database by devising a series of virtual runways at 5,000 feet above the Kansas eastern plains. Then they “landed” the aircraft hundreds of times on those simulated runways.
“It’s similar to how a ski jumper will practice landing in a pool before actually landing on the actual ski slope. We landed at these virtual airports so we could practice the flare maneuvers and make sure the algorithms worked in different crosswind conditions,” Patel added.
At first Garmin tried using GPS alone as its guidance source for the short-final to landing sequence. The results were mixed depending on how good the GPS signal was that day. Although Garmin believed the algorithms worked reliably to account for the GPS error, engineers knew it could be further perfected with the addition of radar altimeter, so that instrument was integrated. In February 2016, Garmin successfully completed its first real runway landing at KIXD. By the spring of 2019, Garmin had completed 800 flights and many, many more landings in the Piper M600SLS.
How Autoland Chooses Its Destination
On the software side, Garmin’s engineers worked on figuring out how to take all the discreet decisions a pilot makes and create a system of prioritization for different combinations of circumstances and conditions. For example, the algorithm considers and then assigns a weight to a whole host of criteria, such as fuel on board, runway length, airspace, real-time weather, terrain, controlled vs uncontrolled airports. It then ranks the choices and selects the most suitable one for landing.
It does all that in .8 of a second or less.
Garmin leaned on its flight test pilots to vet the cascade of decisions and criteria weighting that leads it to the best airport and runway selection. They knew they had it right when the pilots looked at Autoland’s decision and said, “Yeah, that’s the decision I would make.”
According to Eric Tran, senior software engineer who led the design of the routing algorithm, explained that ultimately the individual aircraft manufacturer determines the weighting system, as each airframe has different capabilities and tolerances for weather conditions or runway lengths. For example, a more capable aircraft might be willing to fly through green precipitation returns to reach the most suitable runway, while a light aircraft might fly around them or land somewhere else. The system is also smart enough to realize that if it is currently flying in a precipitation area that is higher than the base tolerance, such as a yellow return, it relaxes that tolerance (since it’s already in it) and continues on, but continues to navigate around yellow or red returns ahead.
Preferred runway length for Autoland is something that is driven by the manufacturers of different aircraft platforms. For the M600SLS, Piper stipulated that 5,000 feet was the ideal minimum runway length, with 4,500 being acceptable if given no other choice. Garmin analysis showed that about 75 percent of U.S. airports equipped with GPS LPV or LNAV/VNAV approaches also have runways that are at least 4,500 feet long.
How Autoland Chooses
After the passenger pushes the Autoland button, the system evaluates all airports around it and determines its route and ultimate destination. It looks to avoid selecting Class B anchor airports, which typically have busy, congested airspace, unless that is the only suitable choice available. It uses all the available weather sources on board, including SiriusXM, FIS-B and/or Iridium datalink weather (but not onboard radar), to help it decide which is most suitable route and landing site. It also considers runway length required, fuel on board, terrain and obstacles, and the availability of a GPS approach with lateral and vertical guidance.
Further, if the aircraft is outside a 20-minute radius of its landing point, it will calculate a forecast based on what it already knows about the weather. For example, if a cell is moving in and forecasted to be on top of the airport, it is capable of predicting that and either choosing a new destination or entering a hold until the cell clears. Throughout Autoland sequence, it will continue to evaluate the weather to reaffirm its runway choice or change based on changing conditions. However, once the aircraft is at the final approach fix, it’s locked in and will continue to a landing. For baro setting, it uses the GPS estimate, and as it gets closer it will use a setting that matches landing pressure altitude, which would be particularly critical in a mountainous area or a location with obstacles. For the M600SLS, the de-ice system is activated and remains on when it detects an OAT below 5 degrees Celsius. It also flies the holding pattern and approach slightly above book speeds because of the unknowns, such as potential airframe ice or wind shear. It does not utilize the onboard radar because of the manipulation of the beam required relative to aircraft altitude and attitude to get the best picture of precipitation ahead.
Likewise, if the system determines the aircraft is too high to begin the approach sequence, it will enter a standard hold to lose altitude and then execute the approach.
During the approach phase, it flies the glidepath using GPS just like the autopilot normally would. On very short final, it blends radar altimeter inputs to decide when to deviate from the glidepath go into vertical speed mode. Garmin tested it in a variety of runway settings such as a sloping runway in Price, Utah, which has a 1.8-degree downslope on Runway 19. It also tested it up to the M600’s demonstrated crosswind to ensure it could crab appropriately and then transition to a smooth and accurate flare-to-touchdown.
After landing, it has a “course” anti-skid function for aircraft installations that have differential braking capability using servos located in the aircraft’s braking system. “We blend the rudder and brake pedals as we slow the aircraft down to help steer it and keep it on the centerline. So, the system differentially applies the brakes and relieve pressure if it feels it slipping. Its only goal is to relieve enough pressure to maintain directional control,” Patel said.
To test the braking capability in the most challenging conditions, Garmin requested airport authorities not to plow the runway or apply de-icing after a major Kansas snowstorm so that they could test the braking system in poor traction conditions. In every instance, Autoland kept the aircraft tracking down the centerline of the runway. After coming to a stop, the system shuts down the engine allowing passengers to make a safe exit.
Human Factors Drive Passenger Interface
One of the most fascinating and well-executed aspects of Autoland is its passenger interface, which the company spent considerable effort to get right. Once Autoland is activated, the normal G3000 displays are replaced with simple visuals accompanied by verbal communications of what is happening and what passengers can expect. The flight displays show the aircraft location along with destination airport, estimated time of arrival, distance to destination airport and fuel remaining. Basic flight information such as airspeed, altitude and heading are labeled in an easy-to-understand format. If passengers wish to communicate with ATC, instructions are provided with the touchscreen interface now functioning as a press-to-talk button. The display helpfully suggests, “speak clearly and slowly” and shows the aircraft’s N-number for reference. Garmin tested every aspect with nonpilots within its human resources team as well as others. They tweaked everything from the calming voice of the verbal instructions to explicit instructions on preparing for landing, and how and when to open the door.
Meanwhile, Autoland squawks 7700 and broadcasts automated announcements on 121.5 as well as tower frequency or CTAF advising that the aircraft is declaring an emergency and executing an Autoland at the selected airport. The system even listens first to ensure it does not step on other transmissions before it broadcasts.
A push of the red button isn’t the only way Autoland can be activated. If the pilot loses consciousness and stops responding to G3000 prompts, the system will execute an emergency descent and prompt the pilot again. If they do respond, Autoland is not activated. If they don’t, Autoland takes over. At any time Autoland can be disarmed by pressing the autopilot disconnect button on the yoke.
According to Bailey Scheel, Garmin’s senior programs manager and systems engineer who led the certification program effort, the program was offered to several aircraft manufacturers, with Piper and Cirrus being the first adopters. At this time, Autoland works with the G3000 NX, Garmin’s current advanced flight deck, with legacy G1000 systems lacking the computing power and capability of supporting Autoland.
The Autoland feature set can be adopted and customized to individual aircraft capabilities and needs. Garmin can foresee a more limited Autoland system for a light piston single. It has said there has been interest up-market in Part 25 aircraft.
“When we started this project, Garmin already had all the pieces – the autopilot, navigation, communication – we just needed to develop a system to have them all work together along with autothrottle and auto-braking,” concluded Carl Wolf, Garmin vice president aviation sales and marketing. “Garmin devoted hundreds of engineers and years of development to make it a reality because we felt it was our mission to save lives. While we hope it is never needed, if it saves one aircraft and its passengers, it will all have been worth it.”
Always on Watch: Autoland Keeps Tabs on Itself
Garmin not only imagined all that the Autoland can do, it was mindful of what it can and can’t do if part of the system is inoperable.
When the pilot powers up G3000 for a flight, the system runs through a normal avionics pre-flight test in which the Autoland components are checked. If an anomaly or failure is detected, a CAS message is posted to inform the pilot. If Autoland or one of the components that it uses is detected to be nonfunctional, the pilot would then have the choice of whether to dispatch on that flight with all or part of Autoland inoperative. If Autoland itself is flagged during the pre-flight test, it’s important to note that the autopilot will continue to function even if Autoland is inoperative.
Where Garmin could partition the functionality and isolate components, such as a brake servo or radar altimeter, they have done so. Thus, if a brake servo fails the self-test, it won’t completely disable the entire Autoland system. However, there are a few essential components that Autoland must have, such as the autothrottle and the pitch servo.
In flight, Autoland is constantly monitoring the health of all its key components required for its function. In the improbable event that the radar altimeter or a brake servo were to fail inflight while Autoland is activated, Autoland will continue with the landing with the capabilities it still has. In this example, it will use GPS for altitude sensing or land without activating brakes.
“We’ve made the system so flexible for OEMs that we haven’t even begun to use all ways it might be configured. While it is fully autonomous, but it could potentially be configured to have a passenger intervene or perform a task if necessary. For example, the system could provide step-by-step instructions to the passenger, even show a picture that says, ‘move this knob or lever and make it look like this’,” said Bailey Scheel, Garmin’s senior programs manager and systems engineer. “That might be a feature made available in a smaller aircraft installation that has less functionality or automation, but still get the benefit out of the technology.”
An example of Autoland’s vast configurability is seen in the Piper M600SLS Halo that incorporates Autoland. If a passenger’s hands or feet touch the controls while Autoland is active and the system senses the controls being moved and posts a message stating, “Keep your hands and feet away from the controls.
Autoland is part of Piper’s newly announced Halo system that bundles safety equipment including autothrottle, electronic stability and underspeed protection, emergency descent mode, level mode, SurfaceWatch and SafeTaxi. At present, autothrottle is only functional for Autoland, but the company plans to offer a fully functional autothrottle by year-end. The M600SLS – the nomenclature stands for safety, luxury and service – also incorporates an updated interior and Piper’s Ultimate Care Program, covering scheduled maintenance and hourly/calendar-based inspections for 5 years.