Blogs

By Jonathan Rupprecht, Esq. for Drone Analyst

This past year, as I attended many drone conferences, I was struck by how eager public safety officials and first responders are to learn how they can incorporate drones into their operations.  Their curiosity is not without good reason.  Small drones with video and infrared cameras are excellent tools for things like situational awareness of critical incidents, search and rescue operations, crime scene processing, and fire damage assessments.

If you are part of one of these outfits and are considering an unmanned aircraft system (UAS) for your operation, then you should know the FAA treats Public Aircraft Operations differently than civilian aircraft operations.  At first glance, the FAA’s requirements for Public Aircraft Operations seem overwhelming, but take heart.  Some of the Federal Aviation Regulations (FARs) do not apply to you, and hopefully this post will clarify and ease what’s required of you.

At the outset I need to say I highly recommend speaking with someone who is knowledgeable about the FAA’s FAR’s, especially if you are seeking to fly your aircraft as a public aircraft operation. The FAA will make sure you follow the very strict requirements to be classified as a public aircraft. If you do not meet the requirements, you will be considered a civil aircraft and could be in violation of multiple regulations.

At the outset, you should know about these three benefits of public aircraft operations status:

Continue reading here: http://droneanalyst.com/2016/01/11/3-things-public-safety-officials-should-know-about-drones/

Having just returned from the vast Consumer Electronic Show (CES 2016), I have a wealth of material to post - including pics, video and soon-to-be opinions and articles on the state of the civilian drone market.

First, however, it might be instructional to describe the current scope of the civilian drone market - in the hopes of helping the public and trade understand the various segments.

For ease of understanding, this article breaks down the civilian drone market into 6 segments. As will be noted, these segments can overlap - especially those above the toy/fpv category.

The diagram above notes 6 basic types of drones. These are separated by size, price, mission, reliability and flexibility - that is, the ability to add various options and equipment. An overview as well as some examples of the breed follows;

1. Toy Quadcopters - This covers models selling from as little as $15 to as much as $450+. The lower priced models can be considered disposable as they will often break or burn out motors within a hour or two of operation. At the higher end you get additional features such as cameras and/or brushless (longer lasting) motors. These models function as novelty items, gifts and learning platforms. Some of the upcoming models are glorified selfie sticks - such as the "flying soccer ball" Flyeye at http://gofleye.com/.
Due to the light weight of most of these units, no FAA or other registration is required. Most are for flying indoors or around your yard or a local small park.
Although these are called "drones", they are usually not autonomous - so the term is more of a marketing trick than truly descriptive. However, as prices for various components drop, it's likely you will see more intelligence built into even these models.

2. FPV/Racing - As with the toy models, these are also not true drones - in fact, most models are purposely very manual in their control and operation. These require various levels of skill to fly - and there are competitive events where users can run courses. As with the toys, FPV pilots get a good eduction in piloting and basic operation. This category is also poised to benefit from advances in Virtual Reality - namely the Occulus Riff headsets. The combination of these technologies will likely give many earthbound humans the ability to truly "fly like the birds". True racing/FPV drones usually sell for $300-$700 complete. Some newer models have GPS and ground sensors to allow for semi-autonmous flight.

3. Consumer Camera Drones - This category has exploded and is providing the impetus (and profits) for Research and Development on more advanced civilian model. Stunning video and photos have become the "killer app" which brought drones into the mainstream. Photography represents an actual "use case" which can be for art, hobby and/or commercial and industrial work. The current range of consumer camera drones (DJI, Yuneec) have proven reliable enough to be used as test platforms for many other applications. These include agricultural, inspection, mapping and more. The low cost of these units (now starting at $499) allow for many new pilots to enter the field. As with any technology, new users bring in new perspectives on the possibilities. When the FAA and other regulations get sorted out, lower cost camera drones are likely to be used for many commercial applications - as they are good enough to do many aerial tasks.

4. Prosumer Camera Drones - such as the Inspire line from DJI, are starting to respond to the need for more serious applications with camera and other upgrades. An infrared camera has been annouced as well as a semi-pro micro 4/3rd standard camera. These models, which sell for $3000-$5000, are likely to serve the expanding markets of film, construction, mapping, agriculture and many more. Although their functions are somewhat similar to the consumer camera drones, they are built using heavier components, have larger payloads and can deal with more wind and other environmental conditions.

5. Consumer, Industrial, Agriculture, NGO and Research Drones - Models from companies such as Matternet and Sensefly are two examples of this breed. Industry leader DJI also just released a $15,000 Agriculture spraying drone. These are models designed as workhorses which will drive much of the future of civilian drones. Current prices range from $5,000 to $20,000+.
As with all other categories, these drones are currently being hamstrung by the lack of clear regulations.
Companies such as MatterNet envision using drones in much of the developing world to deliver medicines, important parts, communications and more. Amazon and others project uses in package delivery although this could be difficult in urban areas. Drones of this size and type are likely to be used in firefighting and other public services.

6. Winged and Vertical Take Off and Landing (VTOL) Drones - are favored for their longer range and greater payload capacities. Large farms, search and rescue and many other applications demand more speed, range and flight times. Models of this type will vary greatly in price - starting in the $10K range and running up to 100's of thousands for very large and sophisticated setups. SenseFly, PrecisionHawk and Autel Robotics (Kestrel Project) are three (of many) companies which are involved in this endeavor.

As we approach out robotic future it's important for the public to understand that drones are a very positive development due to their capabilities. When farming is made more efficient, more people are fed with less waste and pollution. When critical medicine and spare parts are delivered to remote villages, lives are enhanced and saved. Even the simpler task of aerial video allows for people the world over to "travel" to remote places and appreciate the planet - without having to use vast quantities of jet fuel and other energy sources.

There is always a tendency to fear new technology- in the 1960's the whispers started about how computers were going to take all our jobs away. Now many of us have jobs working with computers! The same applies to aerial robotics and robotics as a whole - the future is here and it will be friendlier and more efficient than the past. This will hopefully improve the lives of millions and help us gain the critical energy efficiency we need to cultivate in order to protect our precious planet and resources.

From the AMA blog, just updated http://amablog.modelaircraft.org/amagov/2016/01/11/update-uas-registration-frequently-asked-questions/ 

Q: Why did AMA change its position on registration? Why are they telling members to register now?

A: AMA has not changed its position. We continue to be disappointed with the registration rule and believe it is contrary to Congress’s intent within the Special Rule for Model Aircraft, which is part of the 2012 FAA Modernization and Reform Act. We have been working with our legal counsel and the FAA to find a solution for our members on the registration rule. To date, we have achieved several concessions from the FAA that will help ease this process for our members. However, we want to emphasize that this is not the end of our efforts to protect AMA members from this unnecessary and burdensome regulation. We are continuing to explore all legal and political options available, but these conversations may take time and a long-term solution is unlikely before the February 19 registration deadline. For this reason, we are suggesting that AMA members may wish to take advantage of free registration period, which is available until January 19.

Q: Is the AMA giving up? Why doesn’t the AMA take this matter to court?

A: We are absolutely not giving up. We are continuing to fight for our members and working to protect them from unnecessary and burdensome regulation. AMA is exploring several legal options that would address the registration rule as well. These legal strategies will take time to develop and pursue. We ask for your continued loyalty and patience as we work closely with our legal counsel to find the best path forward.

Q: I am already registered through the AMA, why do I have to register twice?

A: We understand the concern of our members and, while the FAA is open to streamlining the registration for our members, unfortunately the technical issues involved will not be resolved before February 19. Therefore, it will be necessary for current AMA members to register separately with the FAA. For future AMA members, we are working on an agreement with the FAA where new members will be able to opt-in to federal registration via AMA when they join the organization, thereby creating one simple registration.

Q: Am I permitted to fly above 400 feet? What if I had to check a box saying otherwise on the federal registration website?

A: Yes. AMA members who abide by the AMA Safety Code, which permits flights above 400 feet, are still allowed to fly and are protected by the Special Rule for Model Aircraft under the 2012 FAA Modernization and Reform Act. The federal registration box requiring agreement to only fly under 400 feet applies to individuals who are not operating within the safety guidelines of a community-based organization. We have been in discussions with the FAA about this point and the agency has indicted that it will be updating its website in the next week to make clear that this altitude restriction does not apply to AMA members.

Q: Do I need to list both my AMA number and my federal registration number on my aircraft? A: No. AMA has secured a concession from the FAA that allows AMA members to continue using their AMA numbers as the primary identification on his or her aircraft. AMA members are required to still have a hard copy or electronic copy of the federal registration certificate on them while flying.

Q: Do I have to register every aircraft?

A: No. Registration is only required for each operator. AMA members should use their same AMA number on each aircraft they are flying.

Q: Do only drones and multirotor operators need to register?

A: No. Everything over 0.55 lbs (or 250 grams) that uses a ground-control system with a communications link, such as an RC transmitter, is required to register. This includes operators who fly fixed-wing RC aircraft and helicopters, not just multirotors or drones.

Q: I only fly CL or FF, do I need to register?

A: No. If you exclusively fly FF or CL and never plan on using a model that involves a transmitter, then you do not need to register.

Q: Can I fly my large model aircraft? Turbine jets?

A: Yes. The Special Rule for Model Aircraft, which is part of the 2012 FAA Modernization and Reform Act, allows AMA members to operate model aircraft over 55 lbs as long as they are operating in accordance with AMA’s Large Model Aircraft safety program. AMA members can also fly turbine jets provided the operator holds a current AMA Turbine Waiver.

Q: Am I permitted to fly first person view (FPV)? Can I fly at night?

A: Yes. AMA members are still protected by the Special Rule for Model Aircraft, which is part of the 2012 FAA Modernization and Reform Act. As long as AMA members continue to follow AMA’s safety guidelines for these activities, they can continue to fly. The checkbox guidelines listed on the FAA UAS registration website do not apply to individuals who belong to a community-based organization such as the AMA. Q: What happens if I don’t register by February 19?

A: According to the FAA, failure to register an unmanned aircraft may result in regulatory and criminal sanctions. This could include civil penalties of up to $27,500 and criminal penalties that include fines of up to $250,000 and/or imprisonment for up to three years.

Q: What can I do to help? A: You can help by making your voice heard with the FAA. Specifically we are asking all AMA members to submit comments on the FAA’s interim rule on registration. The deadline to submit comments is Friday, January 15. More details can be found here

Introducing the new 100 KM long range Mapping / Video conservationdrones base on Skywalker air frame. With the new power system, the flight duration is more than 3 hours and range over 120 KM with 200 grams payload, typical weight for point and shoot camera such as Canon S 100 GPS on board. The all up weight of the drone is 2.9 KG ready to fly.

About 18 months ago, I've achieved 100 KM with FX 79. But it has never put into practical use in the field. The main reason is with the required batteries on board, if I put payload , such as camera in front, I'll need to add weight at the rear end to get the plane balance. The total weight will be about 3.7 KG and hand launch become very difficult and dangerous. landing a heavy plane at confine area is another challenge. 

I have tried a few combination but non of them meet my requirement until I found the current power system. The major breakthrough is the new Lithium Ion battery pack which rated at 3 C continuous discharge. According to supplier, it use imported LG Lithium Ion battery and the pack is assemble at their partner factory at China. 

The air frame I use is the proven Skywalker 2014 1800 mm three piece wings model. I think other version of Skywalker should give similar performance. The motor is SunnySky 2820 800 kv match with APC 12 x 8 electric propeller. With this set up, you'll need to limit the top end power via Mission Planner to limit the full throttle at about 26 Amps, which give more than enough power for take off and climb to altitude. Once enter cruising mode, the average amps draw is only about 5 to 8 Amps.

The 100 KM attempt took place at my local flying field. It was a sunny Sunday afternoon with some winds. Load the fully charge 4S 21000 mah Lithium Ion battery and some basic set up, it was ready to go. Because of the light weight ( 2.9 KG ) of the plane, I was able to hand launch it with my left hand while using my right hand to hold the radio transmitter. Take off was effortless and after climb to about 60 meter, I switch to RTL mode to test the auto pilot and climb to mission altitude ( 100 meter ). Once everything seems fine I switch to Auto and start the mission.

Three hours and 5 minutes later, It has travelled 117 KM with 29 % battery capacity remaining. As it was getting dark and it actually completed the planned mission, I decited to land. If there was enough time, when push to the limit, I think it can travel another 25 to 30 KM or 40 to 50 minutes. But for general mission flying, it is always good to have about 25 % of reserve capacity for better battery life cycle and prepare for unfavourable weather condition. Another issue need to consider is camera battery life. Most camera will not operate for more than 2 hours without extenal power supply.

The lithium ion battery I used can charge with normal Lipo charger at 1 C rate, but the discharge curve and minimum voltage is difference from Lipo battery that we all familiar with. When high power was loaded, such as take off, the voltage will drop significantly. In my case, the freshly charged battery voltage reading is 16.97 volt. During take off at full power, which draw 25 Amps, the voltage drop to 15.6 volt but once enter cruising mode, it bounce back to 16.4 volt or so. For air plane, it is safe to discharge down to 3 volt per cell or 12 V for 4S pack. If you push to the limit, It can hold up well until 2.8 volt per cell, after which, the voltage will drop fast. You will need to change the low voltage cut setting of your ESC to LOW or change the battery type to NICD/NIMH to avoid premature power cut off by the ESC.

Unlike lithium polymer battery, once the Lithium Ion battery discharge to very low voltage, say 2.4 volt per cell, it can bounce back to about 3 volt if you cut off the power for a while. If you are flying FPV and fight for the last KM to reach home, this technique and the characteristic of lithium ion battery may save your day or at least get it closer to you.

With 100 KM range, the new long range conservationdrone open up the posibility to map or video area that is previously unreachable by normal drones due to unaccessable suitable take off / landing area. It will also suitable to long range river bank ripirian area monitoring or border patrol or any other mission that require long range flight and long flight duration. Flying at 300 meter, it can also map 1500 Ha at about 10 cm per pixel resolution, which make large area mapping possible with a single flight. The camera will need external power.

Below is the specification and parts components for the 100 KM Skywalker conservationdrones. Any experience drone builder can build it with the available information. 

Air frame : Skywalker 2014 1800mm 3 piece wings
Auto pilot : HK Pilot Mega 2.7 master set
Radio control : Futaba T 8 J
Telemetry : HK Pilot 915 air. RFD 900 ground
Motor : SunnySky 2820 800 kv
Servo : Hitec 65 HB
Propeller : APC 12 x 8 Electric
Battery : 4S6P 21000 mah Lithium Ion
On board sensor : Canon S 100 GPS

Here is the link for the tlog of the 117 KM flight. From the tlog, you can view the complete flight mission as well as extract parameter from it. Use it at your own risk as some setting may be difference from user to user. 

https://drive.google.com/file/d/0BwRKGgIOJSELcnp0djliZGs4NkU/view?usp=sharing

Here is the KMZ file 

2016-01-10%2015-15-00.tlog.kmz

Introduction

Flat roofs were originally built in areas with low rainfall. For example, they can be found in the architecture of Egypt, Babylon and Mesopotamia. They appeared in the Central Europe with the discovery of the processing of asphalt strips at the end of the 18th century. In the modern era, the quality of insulation against atmospheric conditions increased so that flat roofs appeared on factory halls as well as highly architecturally appreciated buildings. They are a standard part of all types of buildings. The benefits of such roofs is that they save material, are less labour intensive, provide the option of variable roofing layouts and the potential use of roof areas. However, flat roofs also have disadvantages: the design of the composition of roof layers is complicated, as is the design of the thermal technical evaluation and all principles must be adhered to during implementation. Thermal imaging systems provide the perfect tool to evaluate the technical condition. The thermal imaging system located on the drone provides a fast and economical solution for surveying wide roofs on industrial sites.

Many places on flat roofs can be the source of defects (contact of the roof with the attic, penetration through the cover, inflows, etc.). Problems with leaks can occur in the area, for example, due to unprofessional installation, neglected maintenance or degradation of the surface due to insufficient protection of the surface layers against climatic influences or the end of the service life. In most cases, flat roofs are affected by excessive loading by wind, which causes dynamic shocks, permanent oscillating movement and lifting of unsecured parts. This phenomenon is also the source of the origination of leaks. Water collected on the damaged roof construction negatively affects the durability of roof clad (degradation of multi-layers, leaks).

Thermodiagnostics of flat roofs

A fast and simple way to resolve the problem of the inspecting various flat roofs exists. This method is a thermal imaging system (thermal camera), which due to the thermal difference between a dry place and a place with penetrating moisture, fully identifies discovers where moisture has penetrated. Depending on the construction of the roof, it is possible to see how large an area is affected by moisture, i.e. to localize and state the scope of the problem.

The physics behind the issue of thermal image inspection of problems related to the moisture is based on the fact that the dry part of the roof (not affected by moisture) is cooled and heated slower than the wet part of the roof insulation. Thermal image inspection using a thermal camera during or after sunset (where the roof has been heated throughout the whole time) identifies the moist areas because the dry parts of the roof cool slower than the wet places, which have a higher thermal capacity.

Why diagnostics from the air

Why? It is very easy!

Even when using a thermal camera, which is an excellent diagnostic tool, the question remains how fast and easily it can check the large area of a roof. The way round this problem is to use a pilotless aircraft on which the thermal camera is attached. Such an aircraft can fly above roofs and is not restricted by the number of ladders, elevations or unstable surfaces.

Why to use Workswell AIRvision

The company Workswell developed the Workswell AIRvision system, which combines two camera systems - a visible spectrum camera (for inspection of visible defects found by thermal camera) and a thermal camera for detecting the problems mentioned above. The servicing software enables to remotely switch camera regimes, to record radiometric videos or to make static images in both the visible and infra-red spectrum. The operator sees the objects under the drone in real time or can analyse records and identify damaged areas.

Moreover, Workswell AIRvision, unlike similar systems, enables to manually set the range of temperatures, e.g. in the interval of 15 °C to 25 °C. Even this range of temperatures can be changed during the flight. Without the need to land, the system actual situation can be set for the system and to easily achieve the required thermal sensitivity setting, which is required for the localization of minor moisture defects or the most precise localization of large defects. Traditional photo documentation is required to localise a defect, therefore, the Workswell AIRvision video camera is of sufficient quality to produce a traditional photo. Moreover, the system can be fitted with a GPS sensor for storing information about the position of the drone when creating a record.

The system can measure temperature at the central point, as well as in the local minimum and maximum. Minimum and maximum are localized using a blue (minimum) and red (maximum) cross. This function can also be used to navigate drones because the system automatically shows where the largest potential problem is located.

Produced records are fully radiometric, including videos. After termination of the flight, the records can be evaluated using the Workswell CorePlayer software to produce a measurement report on measurement with the exact localization of the problems. For this reason, a photo is available of all produced thermal images (thermogram).

Individual photos from the radiometric video can be taken and used as separate thermal images (thermograms) to analyse a problem.

More informations about drone thermal imaging system Workswell AIRvision including datasheet and drawings.

This new companion computer from Volta costs just $299 but brings a huge amount of power to Pixhawk-based drones. It's designed to compete with the DJI Manifold at a lower price, and it's open source. OpenCV-based computer vision, streaming HD video, 4G wireless and a 30-second installation with Pixhawk.  Impressive! 

From Ars Technica:

According to the petition for review filed December 24, 2015 at the District of Columbia Court of Appeals, John Taylor is a "model aircraft hobbyist" who owns "one or more small unmanned hobby aircraft" and intends on acquiring more. He does not describe himself as a drone owner in the brief, nor would he confirm himself as such in a conversation with Ars. However, a profile on DIYDrones.com corresponds to his name and location.

In the case, Taylor asks that the appellate court cancel the new registration system that began on December 21. Taylor argues that the new rule is in violation of Section 336 of the FAA Modernization and Reform Act of 2012, which states that the agency may not create new rules if such model aircraft are "flown strictly for hobby or recreational use." If he is indeed a drone owner, then the suit would seem to turn on whether the courts consider drones and traditional model aircraft as distinct from one another.

Other coverage:

http://www.forbes.com/sites/johngoglia/2016/01/04/faa-sued-in-federal-court-over-drone-registration-rules/

http://thehill.com/policy/transportation/264757-feds-sued-over-drone-registration-rules

Research by Ryan Beall and Chaz Henderson:

I did a little more refining of the model from the data previously described/collected from the Pixhawk.  The following are the results of System Identification and PID optimization.

The overall performance of the bank angle controller was suboptimal for our first flights.  It appeared to oscillate in numerous log files (2-3 degrees amplitude with a period of 0.5-1 seconds).  This was indicative of gains that were too high.  A significantly more accurate model was produced by finding a very clean step response in the data as seen below:

This was generated with regression so you can see that it doesn’t fit perfectly.  However, it was clear that this model was a drastic improvement to the previously submitted model.  The regression was also used to model the servo delay (0.04 seconds).  This was more of a black box approach and neglected any cross coupling in the yaw axis (fairly good assumption for this airframe)  The model above is:

As expected, this model is slightly over damped as seen by the pole-zero plot

 This model was then implemented in Simulink to verify performance with PID values used during flight:

Using gains that were close to default (what was used on initial flight), the following was the estimated response for a +-25 degree roll step response:

Roll_P: 0.4

Roll_I: 0.05

Roll_D: 0.05

This performance is very close to what was observed in the flight data logs which is very reassuring that the model is a remarkably close approximation of the aircraft/autopilot system as a whole.

Updating/tuning the PID values based on this approximate model, resulted in the following improved simulated performance

The PID values in the above illustration are as follows and will be used for the next flight evolution.

Roll_P: 0.13

Roll_I: 0.02

Roll_D: 0.04

Note that the P gain was cut in half and the I and D were decreased slightly.  These are pretty low gains compared to most aircraft but this airframe has an absolutely INSANE roll rate and my intuition of cutting the gain in half was well in alignment with the model's prediction.  Hopefully we see verified results on the next test flight... until then...

Fly Safe Fly Fast

-Beall

The video shows localization data collected during a short flight over an IR marker, which is used to estimate the relative location of the copter with respect to the visual landmark. Note that this was a manually flown flight, but the ultimate goal is automation. Detecting visual landmarks/features is a fundamental task in many forms of robot localization and navigation. For example, the Snapdragon Flight includes 4 camera sensors for visual-inertial odometry.

The plot in the video shows the copter’s vision-based position estimation, versus the traditional position estimation. The red data is logged by APM:Copter running on Pixhawk with a 3DR GPS module. The blue data is derived from IR-LOCK sensor data, as it detects a MarkOne Beacon at approximately 50Hz. LidarLite is used for AGL altitude measurements. The presented data looks nice, since it was a fairly tame test. We need to calibrate the lens before we can correctly handle larger pitch/roll angles.

You can think of this as a ‘flipped’ version of the StarGazer indoor robot localization system, where a unique visual landmark is placed on the ceiling. However, the copter localization problem is a bit more tricky, due to the extra degrees of freedom. It can pitch, roll, ascend, etc. So the copter localization estimation also depends on the flight controller’s state estimation. And ideally, all of the data would be fused together.

One of the key advantages of having a uniquely identifiable visual landmark is that it can be used to remove drift in velocity and/or position estimations, which is typically the function of the GPS. This can also be accomplished by developing a local map (i.e., SLAM) …. With the MarkOne Beacon, we can also operate at night, but the video would be even more boring. :) Robust vision performance in variable lighting conditions typically requires some form of IR projection. (see Intel RealSense specs)

The ArduPilot development team is proud to announce the release of the first beta version of the 3.5.0 release of APM:Plane. We think this is going to be a great release and we'd love some feedback before we do the final version.

The biggest changes in this release are:

• switch to new EKF2 kalman filter for attitude and position estimation
• added support for parachutes
• added support for QuadPlanes
• support for 3 new flight boards, the QualComm Flight, the BHAT and the PXFmini
• support for arming on moving platforms

New Kalman Filter

The 3.4 release series was the first where APM:Plane used a Kalman Filter by default for attitude and position estimation. It works very well, but Paul Riseborough has been working hard recently on a new EKF variant which fixes many issues seen with the old estimator. The key improvements are:

• support for separate filters on each IMU for multi-IMU boards (such as the Pixhawk), giving a high degree of redundency
• much better handling of gyro drift estimation, especially on startup
• much faster recovery from attitude estimation errors

After extensive testing of the new EKF code we decided to make it the default for this release. You can still use the old EKF if you want to by setting AHRS_EKF_TYPE to 1, although it is recommended that the new EKF be used for all aircraft.

Parachute Support

This is the first release with support for parachute landings on plane. The configuration and use of a parachute is the same as the existing copter parachute support. See http://copter.ardupilot.com/wiki/parachute/

Note that parachute support is considered experimental in planes.

QuadPlane Support

This release includes support for hybrid plane/multi-rotors called QuadPlanes. More details are available in this blog post: http://diydrones.com/profiles/blogs/quadplane-support-in-apm-plane-3-5-0

Support for 3 new Flight Boards

The porting of ArduPilot to more flight boards continues, with support for 3 new flight boards in this release. They are:

• the BHAT board
• the PXFmini
• the Qualcomm Flight

More information about the list of supported boards is available here: http://dev.ardupilot.com/wiki/supported-autopilot-controller-boards/

Startup on a moving platform

One of the benefits of the new EKF2 estimator is that it allows for rapid estimation of gyro offset without doing a gyro calibration on startup. This makes it possible to startup and arm on a moving platform by setting the INS_GYR_CAL parameter to zero (to disable gyro calibration on boot). This should be a big help when flying off boats.

That is just a taste of all of the improvements in this release. In total the release includes over 1500 patches. Some of the other more significant changes include:

• RPM logging
• new waf build system
• new async accel calibrator
• SITL support for quadplanes
• improved land approach logic
• better rangefinder power control
• ADSB adapter support
• dataflash over mavlink support
• settable main loop rate
• hideable parameters
• improved crash detection logic
• added optional smooth speed weighting for landing
• improved logging for dual-GPS setups
• improvements to multiple RTK GPS drivers
• numerous HAL_Linux improvements
• improved logging of CAM messages
• added support for IMU heaters in HAL_Linux
• support for RCInput over UDP in HAL_Linux
• improved EKF startup checks for GPS accuracy
• added raw IMU logging for all platforms
• added BRD_CAN_ENABLE parameter
• support FlightGear visualisation in SITL
• configurable RGB LED brightness

Many thanks to everyone who contributed to this release! The development team is growing at a fast pace, with 57 people contributing changes over this release cycle.

I'd like to make special mention of Tom Pittenger and Michael du Breuil who have been doing extensive testing of the plane development code, and also contributing a great deal of their own improvements. Thanks!

The upcoming 3.5.0 release of APM:Plane includes support for QuadPlane - a hybrid plane/multi-rotor that allows for high speed long distance flight with vertical takeoff and landing.

The above video shows a converted HobbyKing Firstar 2000 doing a fully autonomous mission, with VTOL takeoff, followed by automatic transition to fixed wing flight and then a VTOL landing. The planes builder, Jack Pittar from CanberraUAV, is shown with the plane below.

The plane itself was fairly easy to build, with the simple additional of rectangular section aluminium arms on the Firstar wings and 4 quad motors.

One issue we found is that the battery we are using (a 3S 4Ah 65C) doesn't handle the 65A needed for vertical takeoff well, with the voltage dropping to 10V on takeoff. After takeoff completes and it transitions to fixed wing flight it quickly recovers. We will be changing the battery setup for future flights.

A second issue is that the wings tend to twist a bit in flight, especially when yaw is commanded. That greatly reduces yaw authority when hovering. We are still thinking about the best ways to prevent wing twist.

The new QuadPlane support is documented at http://plane.ardupilot.com/wiki/quadplane-support and is included in the 3.5.0beta1 release (I will do a separate blog post on that release shortly).

In the near future we expect to add support for other hybrid frame types, including tilt-rotors and fixed wing aircraft with other types of multi-rotor frames attached.

This QuadPlane follows on from our earlier Senior Telemaster QuadPlane which used two Pixhawk flight controllers.

We built the Firstar QuadPlane to test the new code where a single flight controller board running APM:Plane is used, which makes for much simpler operation. We plan on now building a much larger version, with a 50cc petrol motor for fixed wing flight.

This video shows the 250mm quadcopter from my previous post, holding altitude by means of its MaxBotix MB1242 ulatrasonic rangefinder (sonar). The Flip32 flight controller onboard is running a modified version of the Baseflight firmware to support the MB1242.  Original Baseflight requires a barometer for alt-hold, and then fuses it with data from an HC-SR04 sonar if one is available.  So I removed the baro requirement and replaced the HC-SR04 code with code that uses the MB1242.

The real-time plot (made in Python with RealTimePlotter) shows AGL (Above Ground Level) in centimeters (blue) as well as the desired altitude that locks in when I flip a switch on the transmitter (red). The desired altitude starts out with its value from a previous run, which has no effect until I hit the switch, at which point it resets to the current AGL. The bottom plot is the PID (Position / Integral / Derivative) -based correction to the throttle in support of altitude hold, which likewise has no effect until I hit the switch.

Although the vehicle holds low altitude (here, around a meter) pretty well, the values being displayed are quite a bit off from what you expect to see (vehicle is holding AGL at around 100cm, though reported AGL is 60cm, with locked-in AGL at 40cm). At higher altitudes, big transient jumps in the sonar-measured AGL can cause the vehicle to jump slightly. So my next steps are to (1) figure out the discrepancy between the observed behavior and displayed signals; (2) add an Extended Kalman Filter to smoothe out transient jumps in the sonar signal.

Link to full article PDF

So far, the pneumatic connections on our Pitot-static system were considered perfectly airtight. For a change, in this article we examine typical leakage scenarios and their impact on airspeed and altitude measurements. We will focus on leakage solely in the static system; study of the leakage in static pressure piping is important, since at high airspeeds, deviation in static pressure will lead to airspeed measurement errors. Perhaps in the future we will be able to dedicate some time on leakage on the total pressure port as well. In this article, our approach will be to use a lumped parameters leakage model to get preliminary results on the behaviour of our system under leak conditions... [Read more]

The SF40 - upside down :)

The SF40 is a smart sense-and-avoid system that detects obstacles near autonomous vehicles and determines the safest route past them. On-board data analysis reduces the need for complicated processing by an external computer. The built in analytical tools can sense the presence and position of obstacles and locate gaps and spaces with sufficient clearance for safe navigation. Hard-wired alarm outputs make it easy to interface the SF40 with conventional flight controllers and a serial port connection allows for greater interaction when navigation becomes more complex.

How the SF40 works

The SF40 uses a scanning laser rangefinder to measure on a 360 degree disc with a radius of 100m. Collected data is stored in memory and continually refreshed as the laser scans around. The speed of rotation can be set from 1 to 5 revolutions per second corresponding to different measuring resolutions of between 0.03 and 0.25 meters. Time critical functions are managed by an FPGA leaving the 32-bit Arm Cortex-M3 processor to handle the data analysis and monitor system performance and reliability.

Data analysis takes place asynchronously from the data collection so that numerous independent calculations can be carried out very quickly. An analytical tool kit provides the framework for making navigation decisions. Some of the tools run autonomously, such as alarm conditions that are evaluated continuously without the need for intervention by the flight controller. Other tools answer high level navigation questions such as “Is it safe for me to change direction?” or “Which way should I go now?” Tools can be used sequentially and in combination with the situational and internal status information to create sophisticated macros that can handle numerous mission requirements.

Here are some of the navigation tools:

• Alarm zones - autonomously monitors preselected areas to warn of any obstacles
• SearchLight - checks that an area is clear of obstacles before a direction change is made
• Navigator - finds open pathways between obstacles
• Mapper - provides all the distance measurements in a specified region
• VirtualLaserRangeFinder - measures the distance to a target in any direction

Alarm zones

Seven configurable alarms zones can be set within the measuring plane to alert the vehicle when an object gets too close. Each zone can be set with an individualized alarm distance, angular width and aiming direction. For example, one zone could cover 360 degrees around the vehicle at close range to alert when people get too close to the moving parts. A forward looking alarm zone can be used to detect obstacles in the direction of motion. Other alarm zones can check that specific directions are clear of obstructions before course changes are made.

The status of the alarms can be read from the serial port and any two alarms can be linked to the hard-wired outputs. The alarm zone definitions are stored in non-volatile memory making them available immediately when power is applied. Once the SF40 is running the alarms are updated continuously without the need for any external commands.

Two alarm zones are overlaid on the SF40 data:

red = 360 degree safety zone, yellow = forward looking alarm zone

In this case the forward alarm has been triggered by an obstacle.

Checking corridors using SearchLight

The SearchLight tool answers two navigation questions. The first is, “Can I safely change to a new direction?” The second is, “Where is the closest obstacle in that direction?”

SearchLight checks that a corridor is clear before the vehicle changes direction. It can be aimed in any direction on the measuring plane and the beam divergence adjusted to cover the width of the corridor of interest. SearchLight finds the closest obstacle within the corridor and reports its distance and angle from the present flight path.

SearchLight is used to confirm that direction changes are safe before they are made rather than waiting for the alarms to warn that a hazardous situation has arisen after the direction change has been made. Several SearchLights can be used together to triangulate the vehicle’s position between fixed objects and wide beam SearchLights can be used to check for clearance in different quadrants.

The flight controller has responded to the forward alarm by asking SearchLight (green) to look

for space at 45 degrees to the right of the flight path. SearchLight has found obstacles

along the proposed path and warns the flight controller.

Finding clear pathways using Navigator

The Navigator tool answers the question “Which way can I go now?” It examines a specified region and finds the direction of the clearest pathway. The results can be used to direct the vehicle into open space and away from obstacles.

Navigator can be run at any time or activated when a forward looking alarm zone detects an approaching obstacle. The flight controller can configure Navigator with a directional bias so that the recommended escape route is away from known hazards. This can be useful during search and rescue missions when flying close to a cliff face. A bias away from the cliff face would be the safer option.

The flight controller asks Navigator to search in the right front quadrant for an escape route.

There is one safe passage that is clear for 45 m.

The flight controller wants to get further away so it asks Navigator

to find a better corridor in the front left quadrant. Navigator finds a

65m long corridor centered 84 degrees left of the current flight path

Holding station using the VLRF

The virtual laser range finder (VLRF) tool is used to find the distance in any direction on the measuring plane. VLRF can assist with keeping station at a fixed distance from a target or measuring how far away an obstacle is. Any number of VLRFs can be created that aim in different directions. This allows for accurate position holding within a confined space and provides confirmation of GPS location using buildings or other known structures as reference points.

Mapping an area

The Mapping function provides detailed two dimensional maps in polar coordinates of a specified region. These maps can be analyzed by the host controller so that additional measurements or navigation decisions can be made. This is a conventional SLAM mapping feature that won't be needed on most missions.

System capabilities

In addition to collecting and analyzing data from the scanning laser, the SF40 continuously monitors the status of internal systems and the status register can be read by the host controller. System status checks include monitoring the battery voltage, the smooth running of the motor and the correct functioning of the laser. Status flags can be linked to the hard-wired alarms to provide a warning of critical system faults, such as the battery running low.

There is provision to control an external servo motor that can be used to position the SF40 or for any other purposes. The end points of the servo can be trimmed to allow for precise positioning using commands sent through the serial port.

A spare digital input is continuously monitored and its state is reflected in the system status register. This input can be linked to a hard-wired alarm and read using the serial port. This could be used as a landing switch indication.

The SF40 includes a high performance motor controller and efficient power supplies that allows it to run from different battery types and voltages ranging from 6.5 V DC up to 30 V DC. The power consumption is constant at 4.5 Watts.

I hope you enjoyed this brief update :) LD

100 swarming drones from AsTec (now an Intel subsidiary) set a new world record. (The previous record was 30 [correction:the previous record was 50 (thanks Randy)] with a Pixhawk-powered swarm). We're just going to have to wait for the rains to clear and take that record back!

Congrats to the Fleye team, whose innovative PX4-powered single-rotor drone looked great at CES. Here's the Verge report:

The basic premise of the Fleye is simple: take all the spinny, choppy bits of a drone and encase them inside plastic grills and styrofoam, so that it's safer to fly indoors. In fact, it's primarily meant for indoor use: it doesn't go super fast and is controlled with an iPhone or Android phone. There are other drones with covered rotors, but seeing it as a cute little sphere somehow make it more approachable. The main body of the drone is made of the same material you'll find inside bike helmets, and inside that is a single rotor. The air it pushes down is directed by four sets of two fins, independently controlled so that the Fleye can hover and move about the room. It's only good for about 10 minutes of flying time, but it's flying time that is less likely to end in catastrophe because all the moving parts are encased inside.

In our demo, I was able to gently push on the side of the Fleye, and it compensated nicely. You can even just grab the thing in mid-air and turn it sideways, which automatically cuts the power to the rotor. There were a few prototypes around and an engineer constantly tweaking and checking them for bugs and minor hardware issues, but Fleye is confident that it'll be ready for mass production for the fall.

It has a camera that probably won't impress most drone enthusiasts: a 5MP 1080p sensor that can shoot 30fps. The soccer ball-sized drone isn't cheap, either. It will retail for $999, but there are still a few spots and a few days left on the Kickstarter to save a couple hundred bucks on that cost.

Very cool hack in the classic DIY spirit from Hackaday:

Since you’re going to have to be flying your “drones” indoors anyway in the USA, at least in the US Capitol region, you might as well celebrate the one freedom you still have — the freedom to re-flash the firmware!

The Eachine H8 is a typical-looking mini-quadcopter of the kind that sell for under $20. Inside, the whole show is powered by an ARM Cortex-M3 processor, with the programming pins easily visible. Who could resist? [garagedrone] takes you through a step-by-step guide to re-flashing the device with a custom firmware to enable acrobatics, or simply to tweak the throttle-to-engine-speed mapping for the quad. We had no idea folks were doing this.

Spoiler alert: re-flashing the firmware is trivial. Hook up an ARM SWD programmer (like the ST-Link V2) and you’re done. Wow. All you need is firmware.

The firmware comes from [silverxxx], and he’s written all about it on the forum at RCGroups.com. He’s even got thecode up on GitHub if you’re interested in taking a peek. It looks like it’d be fun to start playing around with the control algorithms. Next step, Skynet!

Reading the forum post, it looks like you’ll have to be a little careful to get the right model quad, so look before you leap. But for the price, you can also afford to mess up once. Heck, at that price you could throw away the motors and you’d have a tricked-out ARM dev kit.

And if you insist on hacking everything, you can probably re-purpose a wireless mouse controller to control the thing. Write your own code for the controller and you’ve got an end-to-end open firmware quadcopter for a pittance.

Welcome to Solo, season two.

We designed Solo to evolve. Since we launched, we’ve continued to create and deliver new features that will surprise and empower drone users and moviemakers of every experience level and interest. This latest software release is a big one, and does what we want all our releases to do: It makes every Solo out there a better Solo.

From the beginning we focused on providing a great user experience, helping you quickly and easily capture the exact shots you want, every time you fly. The new Smart Shot functionality in this release — Multipoint cable cam and Follow with Freelook — enhances the platform’s existing Smart Shots. (For the uninitiated, Smart Shots are 3DR’s breakthrough computer-assisted cinema technology that enables even beginners to capture Hollywood-quality aerial video from day one.) We’ll release the new software this spring to all Solo users in a free wireless update.

MULTIPOINT CABLE CAM

Multipoint cable cam extends cable cam’s original design: Now you can create cable shots that run between an unlimited number of keyframes. Set keyframes from multiple points in space, looking in any direction you want. Solo automatically adds spline curves to each keyframe for a smooth approach and exit, so your video will feel professional and polished.

How it works

Fly to the point in space that you want and take your time with the camera to set the perfect keyframe, using your live HD feed to guide you. Press A to save it, then fly to the next one, repeating for as many frames as you want (your flight doesn’t have to be straight line). The Solo companion computer has a photographic memory that takes an internal snapshot of each keyframe, and when you’re done it connects the dots with a virtual cable.

Now engage in three ways:

1) Work the camera freely while Solo flies on the cable from frame to frame;

2) Fly Solo along the cable yourself, while Solo does what no other drone can, automatically handling camera pointing and tilt to capture a smooth narrative from one frame to the next;

3) Or just press “play” and let Solo handle both the flying and the camerawork, perfectly capturing a complex, twisting, tunneling multi-keyframe shot.

The cable cam update will include some interior features, too. Solo will save every cable cam that you create right to your app. Browse by date, location or names you give them.

Return to fly your favorites at any time, or quickly and easily switch between shots at the same location.

Cable cam also has a new enhanced time control feature. Adjust the time of your cable flight in the app to put the camera where you want it, when you want it there. Now that you know exactly where your camera will be, you can turn your attention to directing and synchronizing your shots and scenes.

FOLLOW WITH FREE LOOK

3DR developed the first “follow me” drone. The new Free Look feature turns Follow into an interactive filming experience. In Follow, Solo follows a subject and keeps the camera trained on it wherever it goes. When you enter Free Look the copter keeps following the subject, but you have full control of the camera. Pan and tilt the camera to look around freely while remaining completely confident in the copter’s position and directional heading. It’s like the Hollywood motion control of a camera boom on the back of a truck:

Virtually leash Solo to one subject, then swivel and tilt the camera manually to track even the most spontaneous movements of another subject.

UPDATED SAFETY FEATURES

New software will also include air safety zone information integrated with the Solo app. 3DR was the first drone company to announce a collaboration with an airspace information company — AirMap.io — which will enable the Solo app to pull flight zone information in real time. This helps users make safe and informed decisions about where they fly. Read more about AirMap for Solo here.

We’re also launching another educational software tool, the Solo flight simulator app. The app connects to your controller so that the “sim Solo” you see on your screen responds to your control stick movements just like the copter would in flight. Get a good feel for both flight and camera control without putting your Solo in the air. Plus, you don’t need to connect to Solo to use the simulator; all you need is the app and the controller. Practice flying anywhere.

MADE FOR SOLO ACCESSORIES

The Solo platform will also expand on the hardware side to include an official line of new accessories developed through 3DR’s Made for Solopartnership program. The open gimbal and accessory bays, together with the processing power of Solo, enable any company or individual to develop new hardware for the platform, efforts that 3DR encourages and supports with Made for Solo. New Solo hardware accessories in development include a Kodak 360-degree virtual reality camera; aerial lighting from Fiilex with lights that users can tailor in the Solo app; new Ribcage modified GoPro cameras and camera kits from Backbone with custom weight kits to work with the Solo Gimbal, and more.

Video here: http://tcrn.ch/1OCJHDu

In a recent post I showed how to get live IMU telemetry from a Flip32 flight controller using an inexpensive Bluesmirf Bluetooth modem from Sparkfun. After getting that to work I figured it was time to branch out to some external sensors.

I played around a bit with PX4Flow and LidarLite but thought that for doing something like altitude hold it would be safest to stick with an inexpensive, high-quality ultrasonic sensor like the MaxBotix MB1240, which MaxBotix suggests for  quadcopter applications and other environments with lots of vibration and potential electrical noise. However, based how easy it was to use the Flip32 to read I^2C signals, I wanted an I^2C sensor instead of one that uses TTL, analog voltage, or PWM as the MB1240 does. So I was delighted to find the MB1242, which is the same sensor but with I^2C.

As you can see in the photo below (with the Bluesmirf removed), the same four-wire cable that I used for the Bluesmirf / UART connection was perfect for the I^2C connection, and this time I didn't even have to clip and solder it! (though I did use a Sharpie to color the ground wire black for safety):

Here's a side view showing both the Bluesmirf and I^2C connections:

Having designed and 3D-printed a simple mount for the LidarLite and stuck it firmly to the lower plate with E6000 ("Goop"), I decided to keep that plate attached and print a mate for it to carry the MB1242. Since the MB1242 didn't have enough room for Dee's nuts, I used an additional pair of nylon M3 standoffs to secure the MB1242 to the new mount:

This video shows a test flight of this rig, in the usual "Dronehenge" aerodrome in the Washington and Lee Science Center. As with the IMU hack, I wrote a Python program to display the Bluetooth AGL telemetry in real time. (The Python package I used for the display is on github for anyone who wants to try their hand at this kind of live telemetry visualization.) To support the MB1242 I had to modify the Baseflight firmware to read I^2C from the unit and send it to the UART. This amounted to around 20 lines of C code (send a request byte to the MB1242, then 10 msec later read the two  bytes for AGL in cm). On the laptop, our MSPPG compiler made it easy to generate code that parses the new sonar message from the Bluetooth stream.

Of course, my real goal is to modify the firmware to support altitude-hold based on the sonar AGL. To get started on that I used PyQuadSim, our Python-based quadrotor  simulator, with a simple PID controller to keep the altitude steady. If I can get that working, I will post a link to the modified firmware for others who want to try it.