Embracing uncertainty lets drones fly unknown routes fast

For drones to transform daily life, they will have to move beyond pilots. Unmanned aircraft envisioned for everything from package delivery and rescue work to agricultural survey and military scouting still require pilots or direct operators.

Those human pilots either fly the UAV with a joystick like a remote-control aircraft or guide the drone by entering waypoints, monitoring and navigating the vehicle as it maneuvers. Both approaches limit where UAVs can go and how they can fly, and if drones are going to navigate complex environments such as cities or forests without the aid of skilled pilots, then they will need to learn how to fly themselves in unfamiliar environments.

The Massachusetts Institute of Technology’s Computer Science and Artificial Intelligence Laboratory (CSAIL) has now introduced NanoMap, a tool that lets drones do exactly that.

For people as well as drones, the simplest way to navigate between two points in unfamiliar terrain is to have a map. Unlike a human driver on the road, however, a drone is moving in three dimensions and, outside of a laboratory, that mostly means unmapped spaces.

Mapping the interior of every location or the exact contours of the space between every tree in every forest are impossible tasks. As a result, many navigation systems for drones use a depth sensor and then map the space around the vehicle in real time, allowing the UAV to move as the map is made. That means the drone can only fly as fast as the world ahead of it is mapped.

NanoMap also uses a depth sensor to plot where it is safe for the drone to fly. But rather than stitching together a map of the space around the drone, the sensor takes snapshots of the world ahead of the vehicle and then, if the drone has to maneuver around a sudden obstacle, NanoMap looks back through the previously taken images, piecing together any information about where the drone is going from what it has already seen.

“The problem is that such data can be both unreliable and hard to gather quickly,” says CSAIL. At high speeds, computer-vision algorithms cannot make much sense of their surroundings, it says, forcing the drone to rely on inexact data from its inertial measurement unit, which senses the vehicle’s acceleration and rate of rotation.

“The way NanoMap handles this is that it essentially doesn’t sweat the minor details,” says the lab. “It operates under the assumption that, to avoid an obstacle, you don’t have to take 100 different measurements and find the average to figure out its exact location in space. Instead, you can simply gather enough information to know that the object is in a general area.”

By processing images rapidly, and by accounting for uncertainty, a quadcopter using NanoMap can fly at speeds of up to 20 mph, even through obstacle-rich environments such as a forest or a warehouse.

The research that led to NanoMap was funded as part of DARPA’s Fast Lightweight Autonomy program, which wants to enable drones to operate as people imagine they should: small autonomous robots navigating around obstacles toward human-set objectives, but without humans having to directly supervise or pilot the drones.

The military applications for robot scouts that can find their own way through woods are many. Future uses of this navigation system could create a drone that can fly autonomously through a dense environment to deliver a payload, be it needed medical supplies or a nasty surprise for an unsuspecting foe.

In commercial applications, a drone that can quickly and safely navigate a cluttered warehouse could assist security patrols, or relay needed parts from one side of a complex to another, without fear that it will crash and damage whatever it is carrying.

According to CSAIL, when NanoMap was not modeling uncertainty, and the drone drifted just 5% away from where it was expected to be, the vehicle would crash more than once in every four flights. “When it accounted for uncertainty, the crash rate reduced to 2%,” the lab says.

MRJ Is Meeting Specification And Could Be Made Better

Bradley Perrett

Mitsubishi Aircraft’s MRJ will arrive later than expected, but it may also be better than expected. The company has found it may be able to improve on the specification because aerodynamic and engine performance is on target and time is available for tweaks.

The final assembly plant at Nagoya, Japan, has begun putting together the first two units of the MRJ70 version, which is looking increasingly important for a U.S. market that is still largely bound by union agreements that limit the use of the larger variant, the MRJ90.

Flight testing has addressed the areas of greatest risk, the company says. These include the extremes of center-of-gravity positions, buffet boundaries and performance in stalls. No flutter has been encountered when reaching design speeds.

Testing has also demonstrated that drag of the airframe and the thrust-specific fuel consumption of the type’s Pratt & Whitney PW1200G engine are matching targets. Since the program is still in flight testing, there is time to make minor changes to improve the aircraft, says Yugo Fukuhara, general manager of sales and marketing at Mitsubishi Aircraft. Possible changes could include tightening gaps between panels to lower drag, he suggests. That would lower fuel consumption and increase range, which for the MRJ90 is 3,770 km (2,340 mi.) when carrying the standard load of 88 economy-class passengers and using the type’s design gross weight of 42.8 metric tons (94,400 lb.).

Improving performance beyond the specification depends on winning the usual battle with weight, however. The MRJs that have been built so far, typical of development aircraft, are overweight.

Testing has also revealed that the MRJ’s stalling speeds are lower than expected, offering the possibilities of operating from shorter runways or, from a given field length, at higher weights.

Embraer says its E190-E2 regional jet, now only weeks from certification, is burning less fuel than expected in flight testing. That bodes well for the smaller E175-E2, a direct competitor to the MRJ90.

Because of so-called scope clauses in pilots agreements, the E175-E2 and MRJ90 are too heavy to be used by U.S. regional operators carrying passengers for mainline airlines. In general, the maximum allowable gross weight in such operations is 39 metric tons. Mitsubishi Aircraft offers the MRJ90 with certification paperwork for lower weights of 41 metric tons or 39.6 metric tons, the latter providing an unappealing range of only 2,120 km but still above the scope-clause weight limit.

When Mitsubishi Heavy Industries and partners set up Mitsubishi Aircraft and the MRJ program in 2008, they hoped that by the time the aircraft entered service, then projected for late 2013, the scope clauses would have changed to allow use of larger aircraft. In 2018, they still have not.

“Our strategy will not be dependent on scope-clause relief,” Fukuhara told Aviation Week at the Singapore Airshow, last week. “Our strategy is to introduce the MRJ70 into the U.S. market first,” though the decision is up to the customers. The MRJ70, designed for standard all-economy seating for 76 passengers, is designed for a gross weight of 40.2 metric tons, with a range of 3,740 km. But if given certification paperwork for the scope-clause limit of 39 metric tons, it will still fly 3,090 km. And it may turn out that it can fly a little farther if it can indeed be improved during flight testing.

The Embraer E170, of a similar size to the MRJ70 but using older technology, can fly 3,980 km. But Embraer, while replacing larger versions of the E Jets with the E-2 series, is not preparing a successor to the E170. So the MRJ70 will not have a new-technology competitor.

The MRJ’s two largest customers both fly for U.S. mainline carriers. They are Trans States Airlines, which ordered 50 MRJs in 2011, and SkyWest Airlines, which ordered 100 in 2012. These customers can choose to receive either MRJ90s or MRJ70s, says Fukuhara. It seems increasingly likely that at least some of their aircraft will be MRJ70s.

Development of the MRJ70 is running about a year behind that of the MRJ90. The two MRJ70s that will be used for flight testing of that version are in final assembly. Commonality between the versions is so high that one assembly line handles both.

Despite program delays that have pushed the first MRJ delivery target to mid-2020, the company’s order contracts are still effective, says Fukuhara, implying that the customers cannot walk away from them.

A commercial aircraft development program typically requires 2,500 hr. of flight testing, but the company expects to fly prototype MRJs for 3,000 hr. before first delivery. The extra time creates an opportunity for shaking out more bugs. The prototypes have already demonstrated 99% availability in operations at the program’s main flight-test base at Moses Lake, Washington, but they have hardly been worked as hard as MRJs will be in airline service and, moreover, are under the care of a large and expert engineering team.

Flight testing will last that long because more hours are needed to verify changes. Two redesign efforts were required after the company realized in 2016 that vulnerability to underfloor water ingress and to bomb damage could imperil the MRJ’s airworthiness certification. One change was a reshuffling of avionics in and between the forward and aft avionics bays, the design of which was completed in 2017. The other was changing wiring harnesses. For that, the architecture-level design is complete, but detail design is not, the company says.

Two additional MRJ90s, incorporating the design changes, have been added to the flight-testing program, the company now says; it previously said there would be “about two.” So there will be a total of seven flight-test MRJ90s. Four are currently flying. No showstoppers have been discovered, the company says.

The development effort is now at peak engineering load. As it comes down from that peak, engineers will be reassigned to other aspects of the program, such as product improvement and customer service.

Enders: Partner-Nation Deal Should ‘Reduce the Bleeding’ on A400M

Tony Osborne

The new deal, agreed through a Feb. 7 declaration of intent expected to be signed in the coming months, will rebaseline the program and “further mitigate risks remaining on the A400M program,” Airbus CEO Tom Enders says.

Airbus has so far delivered nearly 60 A400Ms to six of the eight customer nations, but the program has accrued more than €6 billion ($7.4 billion) in losses over the last eight years, with a €2.2 billion charge in 2016 alone because of penalties for the late delivery of aircraft, tactical capabilities and engine problems.

And the company has warned there may be more financial pain to come as it prepares to announce its 2017 results on Feb. 15.

Enders says the company remains committed to the A400M, but the program has suffered from “operational issues” as well as what he called a “flawed contractual setup and insufficient budget.”

The new deal will provide “a good chance to stop or at least reduce the bleeding now and deliver the capabilities our customers need,” he says.

The 2016 charges were the prompt for Enders to restart negotiations with the A400M partner nations—Belgium, France, Germany, Luxembourg, Spain, Turkey and the UK—in an attempt to reduce the financial burden on Airbus. Enders has previously said the company had taken on a “lopsided share” of the risk on the program including—unusually—issues related to the Europrop International TP400 engine.

The contract amendment—agreed by Airbus, the partner nations and the European defense materiel agency, OCCAR—includes a new delivery plan and road map for the aircraft’s tactical capabilities.

Airbus confirmed in January that the production rate for the airlifter will be slowed starting this year. The company plans to deliver 15 aircraft during 2018 and 11 aircraft in 2019, having delivered 19 in 2017. The adjustments will provide a “sustainable future” for the A400M, Airbus says, and should allow more time to find export customers. So far, the only export order has come from Malaysia, whose four aircraft have now all been delivered.

However, last year Airbus did bid the A400M for several competitions, including in New Zealand (AW&ST Feb. 12-25, p. 37).. The status of a long-rumored but unconfirmed order from Egypt is unclear.

The new delivery plan will mean that Belgium, which had planned to take delivery of its first aircraft in 2019, will now receive it in 2020. Belgium and Luxembourg are the only two partner nations that have yet to take delivery of aircraft.

The production slowdown should also suit several of the partner nations that have attempted to offload some of their orders due to budgetary issues. France officially has 50 A400Ms on order but plans for just 25 to be in service by the end of 2025, according to its recently published military program law.

Issues with the delivery of military capabilities have been a frustration for several nations. Germany has criticized the program strongly as it is having to retain several of its elderly C-160 Transall transports in service because of concerns about the integration of defensive aids on the A400M. France and Germany are also planning to purchase tanker versions of the Lockheed Martin C-130J Hercules because of difficulties in developing the A400M’s helicopter-refueling capability.

Airbus Defense and Space CEO Dirk Hoke told Aviation Week at the Dubai Airshow last November that the company has made “good progress” in the development of tactical capabilities in key areas such as dropping supplies and troops, self-defense and helicopter air-to-air refueling.

The company is working with Cobham on the development of a modified refueling pod and hose that can extend farther from the A400M, allowing helicopters to refuel in cleaner, less turbulent air away from the airlifter to reduce the risk of potential collision between tanker and receiver.

The French Air Force has reportedly said aircraft availability stands at 35-40%, although this is largely caused by issues with the TP400 engine that led to an inflight shutdown of an engine in early 2016. Problems with the engine’s power gearbox are being solved more slowly than hoped. The fix, known as Pack 2, addresses problems with the input pinion plug, which was found to be prone to cracking. Europrop International is in the process of completing development of the Pack 2 fix, which is due to be certified during the first quarter of 2018, and engines will be retrofitted during engine shop visits, the engine consortium told Aviation Week.

Questions or Problems? Complete our Feedback Form

Contact an editor or reporter
For Advertising in this newsletter or other digital opportunities contact Jason Washburn
+1 (216) 931-9161 | jason.washburn@informa.com