As you can see from the final round of this season's FormulaPi races, there's still some work to be done to make computer-vision guided cars running on RaspberryPi reliable. But it's a great start (lots of cars completed the course) and I'll be competing in the Spring round, with time to hopefully do better.
From Hackaday:
Ignore the article, watch the video at the top of the page. The article is about some idiot, likely not even a hacker, who bought a drone somewhere and nearly rammed it into a plane. He managed this with concentrated idiocy, intention was not involved. While these idiots are working hard to get our cool toys taken away, researchers elsewhere are answering the question of exactly how much threat a drone poses to an airplane.
https://hackadaycom.files.wordpress.com/2016/11/droneexplode_thumb.png?w=250&h=250 250w, https://hackadaycom.files.wordpress.com/2016/11/droneexplode_thumb.png 714w" sizes="(max-width: 400px) 100vw, 400px" />Airplanes are apparently armored to withstand a strike from an 8lb bird. However, even if in a similar weight class, a drone is not constructed of the same stuff. To understand if this mattered, step one was to exactly model a DJI Phantom and then digitally launch it at various sections of a very expensive airplane.
The next step, apparently, was to put a drone into an air cannon and launch it at an aluminum sheet. The drone explodes quite dramatically. Some people have the best jobs.
The study is still ongoing, but from the little clips seen; the drone loses. Along with the rest of us.
Perhaps the larger problem to think about right now is how to establish if a “drone” has actually been involved in an incident with a passenger aircraft. It seems there are a lot of instances where that claim is dubious.
The level of engineering sophistication here is really impressive. Amazing how fast this industry has matured to smartphone-level professionalism
Huge fun at our Self Racing Cars hackathon today. We'll be having another one in Jan/Feb -- sign up here.
The course could only be completed with computer vision. No GPS indoors..
She did better than some of us pros!
I love this young and smart team, who came out of University of Kansas and are doing very cool work with FPGAs. From sUAS News:
Aerotenna, the leader in radar sensing and SoC flight control technologies for UAVs, has won first prize at the Unmanned Traffic Management (UTM) Preliminary Drone Sense & Avoid technology competition held on November 9 in Syracuse, NY. The competition was launched by a partnership between UTM Convention and HiddenGenius.com to demonstrate drone sense and avoid technologies in an open competition to build public trust.
UTM Convention 2016 is held November 8-10 at the Oncenter in Syracuse, NY. The annual event brings together aerial technology leaders from around the globe, including experts from the FAA and NASA, to explore the latest technical capabilities and discuss collaboration to rework the global airspace in the next era of aviation. The Drone Sense & Avoid technology competition was held during a Google-sponsored reception in front of leading companies and organizations including Google, Intel, Amazon and NASA, as well as drone industry regulators, investors, and most importantly the public.
Sense and avoid is a critical technology that enables safer and wider usage of drones in many applications such as monitoring the safety of bridges, rail lines and other critical infrastructure, delivering packages, and humanitarian missions. The competition obstacle course was designed as a mini replica of environments and scenarios drones would face in the real world, including a miniature bridge, glass obstacle, replica power line, and moving obstacle. More than one thousand in attendance watched as Aerotenna demonstrated its collision avoidance solution based on the μSharp sense-and-avoid radar and the OcPoC SoC FPGA flight controller. Aerotenna successfully completed two collision-free rounds in the fastest time to take home the $12000 first prize.
The development of sense and avoid technology is also critical for the flight of drones beyond visual line-of-sight and into national airspace. Aerotenna recently announced the release of a new miniaturized collision avoidance radar and SoC flight controller for all drone platforms. Radar’s robustness and reliability in all-weather, all-terrain and all lighting conditions make it an indispensible sensor for autonomous smart drones. Until recently, radar has not been considered a practical sensing solution for small UAVs due to size, weight and cost. With a suite of robust yet affordable and easily integrated sensing and processing solutions for all drone platforms, Aerotenna is positioned to be a key solutions provider in the commercial and consumer drone industries.
“Advancements in drone sensing and processing technologies are making autonomous drones a reality, and open competitions such as the UTM Drone Sense & Avoid competition are great venues to show the public just how smart drones have become,” said Dr. Zongbo Wang, CEO of Aerotenna. “The capability of Aerotenna’s collision avoidance radar shows the power of radar as a robust, affordable and easily integrated solution for all drone platforms in the new era of aviation.”
I love both the EMLID Navio 2 autopilot add-on for Raspberry Pi and the Reach RTK GPS. But the latter was a bit hard to use initially, so this news is very welcome: they've now released a major update of their RTK software:
Hello everyone!
After months of work and rewriting the app from the ground up, we are ready to introduce you to the new ReachView generation. Over the past year Reach has become an incredibly popular device and we want the software to be up to par with the hardware. We've used the past year's experience and user feedback to rebuild the app. This update focuses mainly on general stability, user experience and usability. We got shiny looks too
Changes
User Interface
- Completely new beautiful UI
- Dashboard with all important information
- Integrated maps
- Contextual hints
- RTK settings simplified
- Streams now show connection status
- GNSS visibility predictor
- Every field has validity checks
Back-end
- New back-end handles RTKLIB
- Robust performance
Base station mode
- Now Reach can operate both as a rover and a base without mode switching
- Automatic base coordinates by averaging single position or RTK Fix position
- RTCM3 messages rate can be adjusted individually
- All RTCM3 messages now have comments
Bluetooth
- Completely reworked bluetooth
- Supports COM port to Windows
- Works more robust with Android
- Bidirectional communication
- Ability to disable Bluetooth RF
Wi-Fi
- Completely reworked Wi-Fi
- We don't relu on Intel's software
- Add/remove networks from within the app
- Connect to any of known networks
- Switch to hotspot mode
- Device name now can be changed
- Ability to disable Wi-Fi RF
- Hotspot password can be changed now
Memory
- More memory is available for logs
- Base now logs raw data
- Automatic log split
- Rolling logs support
Update process
- Now a dedicated app handles updates
- Update system using packages
- You can see if a new version is available
Bug fixes
- Fixed kernel crash if UART radio talks during boot, leading to device not booting.
- Kernel options for 3G modems compiled into the image
- Fix SPI for robust IMU reading
- New self tests compatible with M8T 3.01 firmware
Known issues and missing features
The app is not suited for multiple users working at once. It will misbehave heavily if open in several browsers simultaneously
RINEX logging. Users can download ubx raw data log files and convert them to Rinex using RTKCONV utility. Soon we will implement on-the-fly conversion and compression. That will save memory and get rid of a separate conversion process
USB connection. USB has been reworked in order to support COM port creation on Windows when connected as a client, but USB features did not make it to this beta. Radio over USB will not work for now.
No progress bar during update process. Update takes about 2 minutes, depending on your internet connection. Please be patient, we know this is hard to endure
After doing the first setup and rebooting, ReachView may experience a problem with a browser's cache. You might see the main chart missing. To fix this, refresh the page. If that does not help, clear browser cache and refresh again.
What's next?
We are actively working on bringing new features to this beta release. Here are some of them:
- Fix issues based on community beta test
- Rinex logging
- USB handling
- Add Galileo support
- RTKLIB update from b16 to b24
- Stop&Go surveying
- Point stakeout
- 3G dongle tutorials
Remember! This is a beta!
Please use with it with caution and care.
How to update?
This release requires reflashing of the device. Please download the new image here1 and follow the reflashing instructions2.
The docs will be updated to the ReachView 2.0 as soon as it is out of beta.
Please do leave feedback, bug reports and any kind of reviews!
We are looking forward to them all.Best regards,
Emlid Team
I'm excited about this, not least because Aero will support the Dronecode stack out of the box. From PC World:
Intel in December will start shipping a fully loaded drone kit to let you do just that, with all the parts including the rotors, software, 3D camera and flight controller.
Intel's Aero Ready to Fly Drone kit will go on sale on the company's website. An Intel spokeperson couldn't immediately provide a price. But it won't be cheap—likely more than $600.
The quadcopter kit has parts that Intel uses to build its own drones. On the company's part, the drone airshows it has organized are getting ambitious: the company has put up 100, and most recently, 500 drones in the sky.
Drones themselves are getting sophisticated. DJI's Phantom has chips and 3D cameras that can navigate safely while avoiding collisions. The Intel Aero Ready to Fly Drone kit has the 3D RealSense camera, which can measure distances and recognize objects and help the drones, when programmed correctly, to fly autonomously to a given destination.
The 3D RealSense camera attaches to a central computer called the Aero Compute Board, which gives the drone its computing horsepower. (Intel also sells the Aero Compute Board separately for $399.) It is powered by a quad-core Atom X7-Z8700 CPU code-named Cherry Trail. It also has LTE, 802.11ac Wi-Fi and a flight controller. The board also has 4GB of LPDDR3 RAM, 16GB of flash storage, a micro-SD slot, a micro-HDMI port, and a wide set of connectors for adapters and breakout boards. It also has the Altera Max 10 FPGA, which can reprogrammed for image recognition, navigation and other deep-learning tasks.
The drone will work with the Airmap software development kit for navigation. Programming will also be required to put the drone in the air. For example, you can use the RealSense SDK to program image recognition for the 3D camera.
An overview on how to build drones from the recent Embedded Linux Conference Europe 2016 is available on Youtube.
The kit will ship in some countries in North America, Europe and Asia. For U.S. buyers Intel has included a caveat relating to government regulations on requiring authorization to fly drones.
"9 Lessons Learned from Attaching a Super Heavy, Custom Rig to a Quadcopter": A very interesting post from Makezine and how to attach things to drones:
Recently, a company called 360.place contacted me about mounting a GoPro VR Rig to an Inspire 1 drone. Getting paid to work with new and interesting technology is pretty much the most fun an engineer can have, so I was thrilled to take on this challenge. After about a month of work designing and building the rig, I had something that the customer was pleased with, however, it definitely took some work to get things right. Here are a few of the lessons I learned on this project, as well as a few that I got right the first time!
When in Doubt, Measure or Weigh
At the start of this project, 360.place decided themselves what type of drone and camera they wanted to use. The good thing is that two of the project conditions were already taken care of, but on the other hand, I had my doubts about whether this drone could lift what we thought was a 5 pound weight. After the vehicle struggled to lift 4 pounds in a test video (not ours) seen below, a 5 pound weight plus whatever I needed to use to attach it seemed like it would be impossible.
Not having my arms “properly calibrated,” 5 pounds seemed like a good assumption after holding it, but when I went back to their office to discuss the project further, I brought a precision scale. Although I thought this might have been the end of the project, the rig actually weighted in at a much more manageable 2 pounds. It was time to start planning
Ideas Come… Eventually
An Early Design Idea
The obvious solution to mount this rig was to run some sort of rod down from the center of the DJI Inspire. This has been done before, but I didn’t see a good way to do it without modifying the quadcopter itself. For better or worse, when I can’t figure something out, it just tends to churn over and over in my head until I have a good solution. Perhaps I should figure out how to charge for these random “unguided brainstorming” sessions. On the other hand, listing “$300: staring off into space for several hours” on an invoice likely wouldn’t go over too well.
The final idea hit me while driving, “Why not mount it to the carbon fiber rods attached to the motors?” I’m not exactly sure how inspiration struck, but having 30 minutes or so where I was entirely free from distraction besides staring at the road really seemed to do the trick. Perhaps taking a break from everything once in a while is a good idea in the creative process. Just remember to note your great ideas down when they come!
Pay in Dollars or in Weight
One thing that struck me while doing this was how expensive carbon fiber and other lightweight components are. The rods attaching the camera rig carrier to the Inspire cost well over $100, and I purchased several components in aluminum that traditionally would be made out of steel. Add to that some machining time to take off extra material, and the costs really start to add up.
Angular Momentum Matters
The whole time I was designing this, I concentrated on weight and keeping the Inspire far away from the VR rig in order to keep the shots as clear as possible. Unfortunately, the farther away from the quadcopter the weight is mounted, the greater the angular momentum acting on it. As seen in the video below, before I reduced the length of the rods, things did not go well:
Or as Archimedes said: “Give me a lever long enough and a fulcrum on which to place it, and I shall move the world.” Perhaps today he’d also add, “…Or death-spiral a quadcopter.”
Test Incrementally
After the first failed experiment, we loaded up the quadcopter incrementally. First we just used the rig with no weight, then a lighter Samsung camera, and finally a weight representing the Omni. There was no real problem this time, but perhaps if we’d tried this procedure the first time, we could have avoided a crash.
Record Your Trials, Know When to Bail
The one upside to this crash was that we did record it from several angles. Besides perhaps being educational for others or even myself pursuing their own filming system, I found it quite entertaining.
One thing that may stick out is that while the pilot tried to go catch the Inspire, you can see me leaving. Though it might not look very “heroic,” having worked on and around machinery for many years, I always try to remember that your hands, eyes, or life isn’t worth whatever piece of equipment you’re working on. When things start to crash, in most cases it’s best just to let it happen.
Practice On Small Drones is Good
During this process, I said that I didn’t want to fly the drone, I’d leave it up to the customer. Probably a good policy, but while I had it to work on, I needed to do some landing gear testing and decided to take it up myself. I had permission, and had been assured that it was OK with the customer, but obviously didn’t want to crash their multi-thousand dollar quadcopter.
To my pleasant surprise though, the controls worked in the same manner as the little Hubsan drones that I’d crashed hundreds of times, and I was able to take off and land with no problem.
Don’t Take Risks with Other People’s Equipment
Though it was easy to fly with no payload, after trying it out myself, I requested to fly the entire rig during our final test. This wasn’t a problem initially, but when I decided to pull in for a landing I was a little off, tilting the quadcopter, and eventually causing it to topple and crash. Fortunately, there was no serious damage, but I was quite embarrassed. The gimbal mount was cracked (re-cracked actually, since I repaired it from the first crash), and one of my custom rod ends pulled out of the carbon fiber.
Not good.
Take All the Tools You Might Need
The good news on this though is that I brought everything I though I would possibly need to the test site. Once I’d examined the defects, within a few minutes I had epoxy drying in the appropriate places, and the drone’s normal pilot went on to film on location in a few days. Here’s some footage that he took:
Overall it was a successful project, and I hope I can help them with other projects in the future. If you’re interested in building your own rig, I put my design and bill of materials up here (with permission) as well as more information on the design process. I invite you to check it out, and hopefully this post helps you realize what went into building it!
Featured Image Photo Credit: PJ Accetturo, 360.place
For those in the SF Bay Area and interested in the DIY autonomous car scene (which I'm also part of), here's the latest on our hackathon this weekend.
Self Driving Racers,
Carl, Joshua and I are looking forward to seeing you all this Sunday (10:00am - 3:00pm) in Berkeley for the meetup/hackathon. Here's some information to help you prepare and otherwise make the most of the event.
It's going to be mostly indoors, in Carl's new workshop (in that happy window between a new space and it being totally filled with equipment). Here's about 1/4 of it, under construction a few weeks ago:
In that space, we're going to have a track. By "track" I mean strips of colored tape, roughly like this (thanks to the Formula Pi team for the inspiration):
If you want to race, bring a ~1/10 scale car/rover (anything under about two feet long) that is capable of autonomous navigation with computer vision. Already got something like that? Great! If not, if you've got a RC car and Raspberry Pi + camera, you can start with the Formula Pi code (forked version here), although please note that that code does assume you're using the ZeroBorg motor controller board with a Raspberry Pi Zero. If you don't have that board, you'll have to modify the code to support whatever motor controller you are using.
I'll have 3-4 Raspberry Pi-based rovers available for people to hack on. They don't have the ZeroBorg boards, so no guarantees that they're ready to roll without modifications. But who knows? Perhaps I'll have them running by then.
As an additional navigation aid, we'll have a ultrawideband indoor positioning system set up. This uses the Pozyx system, so if you've got one of those, you should be able to use that.
What about larger autonomous cars, from go-karts to full-size? If you've got one, bring it for show and tell! We may have access to the parking lot a block away for some slow-speed driving, but no guarantees and definitely no racing.
The famous Carl Bass autonomous monkeymobile will be there. Rides may be offered, although you'll have to fight the monkey for the wheel.
Other stuff that will be provided: Tables, chairs, power, wifi, coffee, pizza. And interesting fun.
Don't worry if you don't have anything to bring/hack on. You're welcome to just hang out and do the meetup thing. Or help others hack their things. This is that brief moment where we can all be n00bs. Someday our cars are supposed to be perfectly safe and autonomous. But for now, let's hack and crash while we still can.
Looking forward to seeing you all on on Sunday.
Chris
Interesting preview of a very powerful new autopilot from Aerotenna:
OcPoC (Octagonal Pilot on Chip) is the SoC FPGA-based open-source flight control platform engineered to bring you greatly enhanced I/O capabilities and processing power that is unparalleled by any other platform of its class. Including the traditional sensor options for common peripherals, OcPoC also expands its input and output capabilities to include fully programmable PWM, PPM and GPIO pins to integrate with a vast number of different sensor additions. It also includes many other standardized connectors for peripherals such as GPS, CSI camera link and SD card. Drone developers can integrate various sensors and have the processing power to not only run ArduPilot but also implement real-time processing of sensor data simultaneously. OcPoC opens the door for drone development to the next level.
OcPoC-Zynq is powered by the Xilinx Zynq processor which combines the flexibility of FPGA architecture with the processing power of ARM, all in one SoC. Along with the I/O expansion, OcPoC provides increased processing power capable of achieving real-time sensor fusion and onboard data processing. This advanced system caters to both the UAV enthusiast that wants a ready-to-fly package and also to programmers and developers wanting a platform to power their ideas.
• First Xilinx SoC FPGA-based flight controller
• FPGA + ARM Cortex A9 dual-core processor
• Over 100 I/Os for sensor integration
• Video streaming and processing capabilities
• Enhanced GPS and IMU sensor packages
• PX4 and APM compatible (www.dronecode.org)
• Open-source hardware and software platform
From CNET
The Federal Aviation Administration last week granted Disney a waiver so it can show off its amazing drone skills in the Orlando and Anaheim parks. Not a company known to wait well, it's already released a video teasing Disney's use of the drones to create what appears to be a floating, rotating Christmas tree.
The clip, posted Monday to its Disney Parks Blog, is brief (less than 30 seconds, and is just a teaser), but shows several people coordinating the drone effort, probably licensed drone pilots as required by the FAA waiver.
In the application to the FAA from October 2015, Disney stated that the drones (which it calls "Flixels"... how adorably Disney is that?) would operate in no-fly zones away from guests in restricted areas. In addition, the filing stated that these Flixels are less than 40 inches in diameter, weigh less than 10 pounds and fly at "six knots groundspeed" or about 7 mph.
The drone-love is nothing new for Disney -- it filed multiple patent applications in 2014, including one for marionettes supported by drones. And this year, the company was granted a patent for a drone with an attached projector.
Keep in mind this is a teaser and not a guarantee that we will see anything this new this holiday season. Much like the Guardians of the Galaxy Mission Breakout or Star Wars-Land, no timeline for these experiences has been announced.
FormulaPi, the remote RaspberryPi-based racing series has started, and it's proving both very interesting and further evidence that vision-based navigation is hard, even on a well-marked track.
Super impressive control implementation to make this work. From Popular Mechanics:
Researchers at the University of Pennsylvania have unveiled the world's smallest self-powered drone, which weighs only 2.5 grams and is the size of a quarter.
The tiny drone is called Piccolissimo, after the Italian word for pocket-sized. The drone comes in two versions: the quarter-sized one, and a slightly larger and heavier one that is steerable.
The drone can fly and steer because both the body and the propeller work together. A tiny motor spins the body while the propeller spins the opposite direction. The propeller is mounted off-center, which is how the drone steers. By precisely changing the speed of the propeller at different points during the rotation of the body, the drone can control the direction of its movement. It's a trick that's been put to use in larger drones before, and is no less clever now that it's smaller.
The drone is built using a 3D-printed frame, a lithium polymer battery, a motor, and a simple control mechanism. These simple parts ensure that the drone is very cheap and easy to build.
"One of the interesting things about the design is that much of the complexity is in the design of the body which is 3D printed," said researcher Mark Yim to Digital Trends. "Since the cost of 3D-printed parts are based on the volume of plastic in the part, and independent of complexity, the flyer is very low-cost."
The researchers hope that their drones could be used in swarms for applications like search and rescue, where hundreds of small Piccolissimos could be used for the same cost as a single large quadcopter.
Source: UPenn via Digital Trends
Congrats to Victor Mayoral and the rest of the Erle team for a successful acquisition by Acutronic. From Victor's story on the company's beginnings and latest deal:
This is a story about how two young brothers of 24 and 21 years old turned a small venture that had nothing more than 3000 € (about 3300 USD) into a multi-million Euro robotics company. The lessons learned in the Erle Robotics journey
Taking off for the first time is tough, always
Erle started in 2012 as a project (ProjectErle Robot) that aimed to create tools for developers in the robotics area motivated by our previous experiences (and struggles). Hardware and software that helped roboticists test and do their work without having to reinvent the wheel every single time they created a new robot — which is pretty much what has happened in robotics over and over. Refer to this article if you’re interested in reading more about the topic.
David and I started with barely 3000 € borrowed from our father, Patxi. At the time, we had no office so we refurbished a room in our father’s house and started it out there. Not many knew what we were up to and the ones that did, didn’t seem to put much interest on it. We were indeed doing something out of the ordinary but more than that, we started noticing that
failure in southern-european countries is unacceptable. The fear of failing kept many away from us from the very beginning.Still, we’ve always been a pair with that grit factor on us so motivated by the growing popularity of the so called drones, we decided to start focusing in aerial robots and slowly designed, prototyped, assembled and tested ourselves the first concepts that later would become the brains for thousands of robots and drones out there. Starting something for the first time is definitely difficult. You probably won’t get it right the first time but if you have right attitude, every misstep will strengthen your vision and ultimately, yourself.
A robotics startup needs cash
After a couple of prototypes that burned our few resources we had the first results. We managed to put together a Linux-based single board base platform for building a wide variety of different robots. It felt great and we could sense that there was potential for doing business with that.
It was time to look for an investment!
Unfortunately, many private investors closed their doors to us. Apps were popular at the time. Everyone thought that Android and iOS were all you had to know in the startup environment. Bootstrapping and getting a minimum viable product over the weekend was common but we were trying to create hardware. Robots. We needed cash.
Robotics startups are totally different beasts from software startups and they need cash from the very beginning.While many investors kept considering our venture too risky, we were lucky enough to get supported by the Basque Government through BIC Araba (previously known as CEIA) and their Ekintzaile program. A grant that helped us iterate and reach a certain maturity point that led us to finally convince the first business angels.
The team is the king
With the first funds available, it was time to grow our team. We hired our father as the finance guy in the company — a person we could trust while being focused in the technical development and commercialization of the technology. We also hired Alex, Iñigo and Irati as our first engineers. Incredibly talented people that helped realize our vision step by step and motivate newcomers. We slowly kept incorporating new members. Lander, Asier and several others followed.
Starting such an adventure required people we could trust so we hired carefully but mistakes were also part of the path. If there’s something we learnt out of these years of work is that
in a startup one needs to be prepared to hire but also to fire.Stay away from individuals that look solely for their self-benefit and growth while getting away from the company’s core values. These last years taught us that understanding your team is as relevant as having a clear vision of what’s ahead in the business landscape. The team is the king and those that fail at accepting that, don’t fit.
The secret sauce … a.k.a. the hidden founders
Many startup stories talk solely about the founders: their inspiring vision, their leadership, their hiccups, … and while much of this remains true for a great majority of success cases, there’s something as relevant that rarely gets mentioned.
This secret sauce is present in every single startup in the world. I call this sauce the “hidden founders”.We had two ingredients in our recipe, each one of them as relevant as the founders:
- The first hidden founder was Patxi Mayoral Pizarroso and as you probably guessed, he’s the father of the founders. Patxi, helped from the very beginning in every angle. Providing business advice, technological advice, management advice … he took part in the conception of the company and he was there in every single initial test. As the company grew, Patxi took a financial and admin position which helped us focus enormously. Patxi was the glue of Erle Robotics. He helped figure out the best outcome for every problem. Founders have issues between them. This is a well known fact and Patxi found a way to always support us all (which was extremely difficult in his situation). He supported the vision from the background and did all that work that’s never been mentioned. We wouldn’t be where we are today without Patxi.
- The second hidden founder is Alejandro Hernández Cordero. Alex and I knew each other from college and he could’ve joined the team as a founder but out of personal reasons, decided to join later and became the first employee of the company. Despite the fact of his non-founder condition, his commitment has been remarkable and he’s been there in every single relevant scenario providing much more than technical support. He’s an outstanding engineer with worldwide experience in the robotics area. He’s brilliant, honest and well-spoken. He helped us lead the team and create a solid group. Every startup needs an Alex and chances are that you might already know your Alex before even founding.
Celebrate milestones
As it’s already been reported in several places, Erle Robotics has officially been acquired by Acutronic Robotics, called to become a leading robotics firm focused on the next-generation robotic solutions.
The new company will have three business lines: it will be responsible for the development and commercial activities related to H-ROS; robot and drone hardware products will continue to be sold under the Erle brand and engineering services will be offered to corporate and government clients.
One thing that entrepreneurs often forget is that being focused is as relevant as knowing when to celebrate. I strongly believe we’ve failed at this. Our focus has been constant and I don’t remember the last time we celebrated something as a group. Bülent Atlan a great technologist and an even better advisor once shared a few of his lessons with me. I couldn’t agree more with:
If you work hard, party hard.Spanish people have (at least) two last names
No kidding, in fact we’ve got more than two but we just use the first two. Let me take my brother’s name as an example. His full name is David Mayoral Vilches. Now, let me analyze his name:
- David: First name.
- Mayoral: First last name, generally comes from the father.
- Vilches: Second last name, generally, it comes from the mother.
I know this is silly but I’m sick of this. Specially given the fact that Spain has one of the oldest and most influential cultures in the world. So either because you’d like to be knowledgeable from a cultural perspective or just because you want to be respectful, keep in mind that:
it’s not common nowadays for spaniards to have a middle name and likely, that thing after his first name is his last name.Getting called “Mr. Vilches” feels weird (the right form should be “Mr. Mayoral” although “Mr. Mayoral Vilches” would be a total milestone).
The future ahead
Erle Robotics has nowadays been acquired by a multinational and our team has been placed at the center of a new robotics division that will bring many of our ideas to industry. We’re extremely excited about this chance to push our vision around hardware for robots further (summarized in H-ROS) and contribute towards a future where interacting with robots is not restricted to a few individuals with high technical (and mathematical) skills but to a great majority of the users out there.
We’ve been extremely lucky over the last years to hire more experienced and talented profiles like Carlos Uraga Pastor (who took over the CEO position in Erle). Carlos helped us reach the maturity we needed with an extremely committed attitude and a skillset similar to that of many founders. Carlos and everyone else will remain with us in future adventures and we are all extremely excited about leading the changes of what’s coming next in robotics since we truly believe that:
The best way to predict the future is to invent it — Alan Kay
I’d like to take this chance to thank those individuals and organizations who supported us during this period, particularly thanks to my family (that’s you mom!), my girlfriend Keila, BIC Araba (previously known as CEIA) for their support, IE Business School for an amazing training, TechFounders program for helping maturing our ideas and connecting us with the right networks and the overall ecosystem of the Basque Country and Spain for their constant support and aid.
Special thanks to our team at Erle Robotics. The best prize has been having the chance to work with you. We couldn’t ask for a better group.
Interesting glimpse of the single-board design. Impressive that he get it back together again afterwards!
This render from Fusion 360 features the Hooligan 1000, with an additional thruster and rotors set at 30-degree pitch for optimal thrust. The final design may change based on the results of Autodesk Dreamcatcher. / Credit: Eli D'Elia, Autodesk
From Core77:
Someday in the not-too-distant future, a drone designed by a computer will be flown by a computer. That's Eli D'Elia's dream.
It's a dream he's working to make a reality by partnering with Autodesk to design the Hooligan 1000, among the first entrants in the newly emerging 1,000 mm drone racing class. But racing is just the beginning for D'Elia, a product designer, roboticist, and professional drone operator who has helped to pioneer drone racing as we know it. For him, it's more important that the Hooligan be durable, practical, and easily adapted for use in industry and agriculture, possibly search and rescue, and eventually, mapping and exploration. He envisions a future where drones move not only through the air, but through water and outer space as well. "We're really just in the Kitty Hawk stage of drone development," says D'Elia. "Competition through sports is a great way to test out ideas and get rid of the bad ones."
Letting the Computer Do the Work
"AI-grown, AI-flown," is how D'Elia summarizes his goal. To that end, D'Elia and his partners at Autodesk, Taylor Stein and Daniele Grandi, will use the Autodesk Dreamcatcher generative design system to design the drone's chassis. With generative design, the designer doesn't come up with the actual design, but instead inputs a set of requirements and lets the computer go to work. While still in its infancy, it has already been used to create an experimental automobile chassis and components for commercial aircraft.
After the design is finalized, D'Elia will start working on the AI control system that will steer the craft using 3D photogrammetry and the NVIDIA Jetson TX1 processor. He expects the AI to be able to learn a given racecourse after several test flights, then fly it both with and without human control. But he also wants it to have situational awareness, the ability to identify objects in the air like birds and other drones, and respond appropriately. "We're putting an additional thruster in there to provide a turbo boost at the end of the race," D'Elia says. "I'd like the drone to be able to tell if there's another drone close to it so it knows when to kick that thruster in for some extra speed."
Drones Far and Near
What D'Elia wants to do and what the Federal Aviation Administration (FAA) will permit are two different things, of course. According to current FAA regulations, all drones in the U.S. must have a pilot at the controls and be within the pilot's line of sight. And it may be years before autonomous drones can be flown in civilian airspace simply due to the danger of a drone hurting people, damaging property, or interfering with air flight. "They'll need to do things like employ redundant flight control systems so that, if a propeller goes out, it won't fall out of the sky," says D'Elia.
But there's no question that many people and companies see a future for drone autonomy, both near and far. Amazon is testing delivery by drone in the UK, where more lax regulations permit drones to be operated beyond the pilot's line of sight. Meanwhile, the startup, Zipline, is already using autonomous winged drones to deliver emergency medical supplies to distant communities in rural Rwanda. And companies in the U.S. are petitioning the FAA for waivers to the line-of-sight rule. The first such waiver was granted in August to PrecisionHawk, an aerial data analysis company, after they provided a year's worth of drone flight safety data.
For D'Elia, all of this is interesting, but the truly exciting possibilities lie further afield. He notes that there's already an XPRIZE competition underway to design a drone that can map the ocean floor by itself. D'Elia thinks it's only a matter of time before similar efforts begin to explore outer space.
In the Field and in the Lab
When he's not working on the Hooligan 1000 designs, you'll often find D'Elia flying drones in the vineyards of Northern California. The company he started, Eagle Eye Metrics, deploys drones to help farmers, generally vintners, monitor their crops and map their fields. Using a NDVI camera, his drones capture infrared frequencies that can reveal information about soil quality, pest infestation, and overall plant health. On foot, this kind of necessary recon can take farmers a week. D'Elia and his drones can do it in less than an hour.
When he's not designing drones, Eli D'Elia performs drone-based aerial mapping for farmers in Northern California through his company, Eagle Eye Metrics. / Credit: Eli D'Elia
D'Elia has always had an interest in both the sport and the practical use of robotics. Long before BattleBots hit the airwaves, he was part of an amateur robotics community in the San Francisco Bay Area in the '90s. They would build their robots during the week, then put them into the arena to do battle on Saturday night. "The whole idea was to one-up whatever the other guy's robot did. If he clobbered you, you hit him with a projectile. If he fired at you, you hit him back with a flamethrower," D'Elia says.
When drones hit the hobbyist market in the mid-2000s, D'Elia was immediately interested. However, he and his frequent collaborator, Marque Cornblatt, saw one big problem: drones were too fragile. "You could spend two weeks and $1000 building a drone, but you can make one mistake and crash it in ten seconds and you're back to square one," D'Elia says.
Under the auspices of their own organization, the Aerial Sports League, D'Elia and Cornblatt designed the Hiro, a drone with a monocoque frame made from a lightweight, super-strong polycarbonate used in aerospace and military applications. The design was fireproof, waterproof, and sturdy enough to withstand direct hits from a baseball bat and even a shotgun blast. It was among the first successful drone projects on Kickstarter.
D'Elia and his Autodesk collaborators are currently finalizing design parameters for the Hooligan using Autodesk® Fusion 360™ software, specifying requirements for mounting connections and access ports. When completed, they'll feed the requirements into Dreamcatcher, which will take several weeks to generate designs for the chassis. They'll then print that design in polycarbonate, assemble the components, and take it out for a test flight. Then they'll begin working on the pilot AI with help from NVIDIA.
D'Elia remains upbeat about the near future not only for his project, but for drones overall. "Sometimes you hear people say that all this amazing stuff is going to happen 'in the future,' but you have to remember—that's like three to five years from now," he says. "Think how far self-driving cars have come in five years. This stuff is coming and it's coming fast."
Eli D'Elia and Taylor Stein will be speaking about designing racing drones using Autodesk Fusion 360 software and Flow Design wind-tunnel simulator as part of Autodesk University 2016 in Las Vegas, November 15-17. Learn more and register today.
The wait is over! We are proud to introduce the next generation 3DR autopilot, Pixhawk Mini. Pixhawk Mini is an upgraded Pixhawk designed in collaboration with HobbyKing and optimized to run the Dronecode PX4 firmware stack and QGroundControl multi-platform ground station (Windows, Mac, Linux, Android, iOS).
For just $199, Pixhawk Mini includes autopilot, GPS, and all the cables and connectors needed to get started building DIY quads, planes, rovers, and boats.
What's improved over Pixhawk 1?
One third the size--dimensions reduced from 50 mm x 81.5 mm x15.5 mm to 38 mm x 43 mm x 12 mm. Smaller airframes can now operate autonomously without making sacrifices for the Pixhawk footprint.
Rev 3 STM32 processor allow for full utilization of 2MB flash memory. Pixhawk Mini operates at only 50% compute capacity, 40 percentage points lower than the original Pixhawk. There is significantly more overhead available to run custom code.
Improved sensors, including both primary and secondary IMU (MPU9250 and ICM20608, respectively), lead to much better vibration handling and increased reliability.
GPS module included--Neo M8N with quad-constellation support and upgraded HMC5983 compass.
Micro JST connectors replace DF-13. We can all breath a sigh of relief.
Integrated piezo speaker and safety switch
What's improved over Pixfalcon?
Again, improved sensors, including both primary and secondary IMU (MPU9250 and ICM20608 respectively) for much better vibration handling and increased reliability.
Dedicated CAN port for UAVCAN applications.
Includes 8-channel servo output board for planes and other vehicles requiring powered PWM output.
Includes I2C breakout board for a total of 5 I2C connections.
Pixhawk Mini features an advanced processor and sensor technology from ST Microelectronics® and a NuttX real-time operating system, delivering incredible performance, flexibility, and reliability for controlling any autonomous vehicle.
SPECIFICATIONS
Main Processor: STM32F427 Rev 3
IO Processor: STM32F103
Accel/Gyro/Mag: MPU9250
Accel/Gyro: ICM20608
Barometer: MS5611
Dimensions: 38x43x12mm
Weight: 15.8g
GPS Module: ublox Neo-M8N GPS/GLONASS receiver; integrated magnetometer HMC5983
Dimensions: 37x37x12mm
Weight: 22.4g
Interface
1 x UART Serial Port (for GPS)
Spektrum DSM/DSM2/DSM-X® Satellite Compatible RC input
Futaba S BUS® Compatible RC input
PPM Sum Signal RC Input
I2C (for digital sensors)
CAN (for digital motor control with compatible controllers)
ADC (for analog sensors)
Micro USB Port
What’s Included?
Pixhawk Mini Flight Controller
GPS with uBlox M8N module with
Concurrent reception of up to 3 GNSS (GPS, Galileo, GLONASS, BeiDou)
Industry leading –167 dBm navigation sensitivity
Security and integrity protection
Supports all satellite augmentation systems
Advanced jamming and spoofing detection
Product variants to meet performance and cost requirements
Backward compatible with NEO‑7 and NEO‑6 families
Integrated Power Module (up to 6s batteries) and power distribution board for quadcopters
8-channel servo output board for planes and other vehicles requiring powered PWM output.
Cables
4 pin I2C cable and breakout board
6 pin GPS+Compass cable
6 to 6/4 ‘Y’ adapter for additional I2C devices
4 JST to 6 DF13 cable for legacy telemetry radios
External safety switch cable
RCIN cable for PPM/SBUS input
8 channel RC output cable
6 pin power cable for included Power Distribution Board
OPTIONAL ACCESSORIES
All available here
From Victor Mayoral:
Announcing alpha support for the PX4 flight stack in a path towards drones that speak ROS natively.
The drones field is an interesting one to analyze from a robotics perspective. While capable flying robots are reasonably new, RC-hobbyists have been around for a much longer time building flying machines developing communities around the so called flight stacks or software autopilots.
Among these, there’re popular options such as the Paparazzi, the APM (commonly known as ardupilot) or the PX4. These autopilots matured to the point of acquiring autonomous capabilities and turning these flying machines into actual drones. Many of these open source flight stacks provide a general codebase for building basic drone behaviors however modifications are generally needed when one has the intention of tackling traditional problems in robotics such as navigation, mapping, obstacle avoidance and so on. These modifications are not straightforward when performed directly in the autopilot code thereby, in an attempt to enhance (or sometimes just simplify) the capabilities of autopilots, abstraction layers such as DroneKit started appearing.
For a roboticist however, the common language is the Robot Operating System (ROS). Getting ROS to talk to these flight stacks natively would require a decent amount of resources and effort thereby, generally, roboticists use a bridge such as the mavros ROS package to talk to the flight stacks.
We at Erle Robotics have been offering services with flying robots using such architecture but we’ve always wondered what would be the path towards a ROS-native drone. In order to explore this possibility we’ve added support for the PX4 Pro flight stack.
Supporting the PX4 Pro flight stack
The PX4 Pro drone autopilot is an open source flight control solution for drones that can “fly anything from a racing to a cargo drone — be it a multi copter, plane or VTOL”. PX4 has been built with a philosophy similar to ROS, composed by different software blocks where each one of these modules communicates using a publish/subscribe architecture (currently, a simplified pub/sub middleware called uORB).
In an internal attempt to research the path of getting ROS-native flight stacks and to open up this work to the community I’m happy to announce official alpha support for the PX4 Pro in all our products meant for developers such as the PXFmini, Erle-Brain 2 or Erle-Copter. Our team has put together a new set of Operating System images for our products that will help you switch between flight stacks easily.
To install PX4 Pro, just type the following:
sudo apt-get purge -y apm-* # e.g.: apm-copter-erlebrain
sudo apt-get update
sudo apt-get install px4-erle-robotics
ROS-native flight stacks
Using the PX4 Pro flight stack as a starting point, our team will be dedicating resources to prototype the concept of a drone autopilot that speaks ROS natively, that is, that uses ROS nodes to abstract each submodule within the autopilot’s logic (attitude estimator, position control, navigator, …) and ROS topics/services to communicate with the rest of the blocks within the autopilot.
Ultimately, this initiative should deliver a software autopilot capable of creating a variety of drones that merges nicely with all the traditional ROS interfaces that roboticists have been building for over a decade now.
If you’re interested in participating with this initiative, reach us out.
Very nice use of the Microsoft Azure cloud AI APIs to make a AR.Drone do some pretty magic stuff. Excerpt from the full O'Reilly post, which has the full details and code:
The Azure Face API is powerful and simple to use. You can upload pictures of your friends and it will identify them. It will also guess age and gender, both functions of which I found to be surprisingly accurate. The latency is around 200 milliseconds, and it costs $1.50 per 1,000 predictions, which feels completely reasonable for this application. See below for my code that sends an image and does face recognition.
I used the excellent ImageMagick library to annotate the faces in my PNGs. There are a lot of possible extensions at this point—for example, there is anemotion API that can determine the emotion of faces.Running speech recognition to drive the drone
The trickiest part about doing speech recognition was not the speech recognition itself, but streaming audio from a webpage to my local server in the format Microsoft’s Speech API wants, so that ends up being the bulk of the code. Once you’ve got the audio saved with one channel and the right sample frequency, the API works great and is extremely easy to use. It costs $4 per 1,000 requests, so for hobby applications, it’s basically free.
RecordRTC has a great library, and it’s a good starting point for doing client-side web audio recording. On the client side, we can add code to save the audio file:
I used the FFmpeg utility to downsample the audio and combine it into one channel for uploading to Microsoft:
While we’re at it, we might as well use Microsoft’s text-to-speech API so the drone can talk back to us!Autonomous search paths
I used the ardrone-autonomy library to map out autonomous search paths for my drone. After crashing my drone into the furniture and houseplants one too many times in my livingroom, my wife nicely suggested I move my project to my garage, where there is less to break—but there isn’t much room to maneuver (see Figure 3).
Figure 3. Flying the drone in my “lab.” Source: Lukas Biewald.
When I get a bigger lab space, I’ll work more on smart searching algorithms, but for now I’ll just have my drone take off and rotate, looking for my friends and enemies:
From Sparkfun:
Let’s savor our final look back at AVC 2016: It’s time for the final recap video for the Power Racing Series and Autonomous Power Racing Series events! If you enjoy watching full-grown adults in costumes strap themselves to souped-up Power Wheels (plus shopping carts, Batmobiles…and other “vehicles”) – human-controlled and decidedly otherwise – you’ve come to the right place.
