Jordan BunkerSophi Kravitz @Piotr Esden-Tempski welcome!
Piotr Esden-Tempski I asked Felix Ruess (the Paparazzi project maintainer) to join us but unfortunately he is not available this week. Same thing with Gareth McMullin from Blackshpere Technologies aka Black Magic Debug...
Sophi Kravitz Could you introduce yourself and tell us a little about Paparazzi (for those who didn't watch the video)
Piotr Esden-Tempski Hi @Sophi Kravitz thank you for having me. :)
Piotr Esden-Tempski Hi I am Piotr Esden-Tempski, I am the founder of 1BitSquared an open source Hardware and Software development company with the focus on Unmanned Autonomous Aircraft and Embedded systems. We develop and manufacture hardware based on the needs of open source projects like Paparazzi UAV (paparazziuav.org), Black Magic Debug (black-magic.org), GDB, libopencm3 (libopencm3.org) or 1Bitsy (1bitsy.org). We also provide consulting services to companies that are interested in contribution to open source software and hardware.
Piotr Esden-Tempski I have been involved with UAV since about 2006 and wrote my computer science diploma thesis at Ascending Technologies developing autonomous landing platform hardware and software. Since my thesis I continued to work on elements of UAV systems that were in need of open hardware solutions, founding Open-BLDC motor controller project and later joining the Paparazzi UAV project. I helped developing the Lisa/L autopilot for Paparazzi UAV that was to my knowledge the first commercially available autopilot using the STM32F1 microcontroller. And a series of revisions of the Lisa/M autopilot that continued pushing the reliability and computational power of hard realtime autopilots as well as the use of low cost cell phone market MEMS sensors. (we were using ADXR mems Gyroscopes before that) I have also developed Lisa/S that is the smallest open source full outdoor autonomy capable autopilot.
msm1089 Hi all, how's it going?
Piotr Esden-Tempski Through my involvement with open source ARM Cortex based UAV autopilots as well as libopencm3 low-level hardware library for ARM-Cortex microcontrollers, I also got involved with the Black Magic Debug JTAG/SWD debugger project with built in GDB server. 1BitSquared took over the hardware development, manufacturing and distribution of the Black Magic Probe native hardware in 2016 and we work with Blacksphere Technologies on expanding the programming and debug platform further by bringing it to more microcontrollers and keeping Black Magic Debug stable and future proof.
Piotr Esden-Tempski Ok that is the condensed background story of mine. :)
johnowhitaker Cool stuff!!
Sophi Kravitz what is Parapazzi? A little more detail...
Piotr Esden-Tempski We also have @Rich Burton in the chat, he is the mastermind behind the Hooperfly rotorcraft UAV platform. He is also the developer of the Robot Commander agent software for the Paparazzi platform. https://github.com/paparazzi/flyingrobotcommander and is building fully autonomous self directly rotorcraft swarms based on the Paparazzi UAV platform.
Piotr Esden-Tempski Regarding paparazzi uav
Sophi Kravitz welcome new people, sheet for discussion is here: https://docs.google.com/spreadsheets/d/1r5NrwLhG-H7lLR3-Vn5lqVvnJ3yNiWyhCvSNLfwARoA/edit#gid=0
Piotr Esden-Tempski Paparazzi UAV is an open source uav platform for autonomous uav
Rich Burton Paparazzi UAV (Unmanned Aerial Vehicle) is an open-source drone hardware and software project encompassing autopilot systems and ground station software for multicopters/multirotors, fixed-wing, helicopters and hybrid aircraft that was founded in 2003. Paparazzi UAV was designed with autonomous flight as the primary focus and manual flying as the secondary. From the beginning it was designed with portability in mind and the ability to control multiple aircraft within the same system. Paparazzi features a dynamic flight plan system that is defined by mission states and using way points as “variables”. This makes it easy to create very complex fully automated missions without the operators intervention.
Piotr Esden-Tempski It was developed by researchers from ENAC which is the French school for FAA
Piotr Esden-Tempski ahh right that is a good summary that @Rich Burton posted there :)
Sophi Kravitz Thanks!
Sophi Kravitz Can we start with the first question?
Piotr Esden-Tempski Sure go ahead. :)
Sophi Kravitz From @ken.do : In terms of weight to power and power consumption ratios; are brushed motors a good choice for really small multirotors? If so, up to what size?
Piotr Esden-Tempski Coreless brushed motors are quite nice as they are small and easy to drive in small sizes, and you do not need a motor controller. Still if you can find a brushless motor/motor controller combination for the size of aircraft you are building it will still result in a better efficiency of the system. It really depends on the components you are willing to spend money on. Brushed motor solutions are mostly just cheaper than the equivalent brushless solutions for ultra small aircraft.
Sophi Kravitz Why don't you need a controller for brushed motors?
Piotr Esden-Tempski Ok that was not fully accurate on my end. You still need a controller but those are much simpler.
Piotr Esden-Tempski in the simplest solution for a brushed motor you only need one mosfet switch with PWM drive to achieve control
Piotr Esden-Tempski for a brushless motor you basically have to build an electronic variable frequency inverter
Piotr Esden-Tempski but thanks to the advancement of power electronics stemming from handheld device minutarization
Piotr Esden-Tempski the brushless motor controllers can achieve incredible power to size ratios
Piotr Esden-Tempski Till a few years ago I would have agreed that coreless brushed motors are still probably the best solution for sizes under 12cm motor to motor distance or ultra small flying wing aircraft
msm1089 How much smaller can motors get?
Piotr Esden-Tempski but this is not really true any more
Sophi Kravitz so brushed motors use simple PWM, what signal is typical for brushless?
Piotr Esden-Tempski Yes brushed motors use PWM directly to govern the amount of power to provide to the motor. Thus the single switch. For brushless motors you need 3 half bridges that control the three phases that are connected to the motor controller.
Piotr Esden-Tempski Basically the thing that the brushes and the commutator do in the brushed motor needs to be implemented in electronic hardware.
Piotr Esden-Tempski and advance the electromagnetic field in sync with the mechanical rotation
Piotr Esden-Tempski this is more complicated but gives additional opportunities for more efficiency
Piotr Esden-Tempski also brushless motors can be water proof
Sophi Kravitz Thanks Piotr! Good explanation!
Sophi Kravitz Looks like we're covering on some level, Q#2: The advancement of hardware components in the last 2-3 years. Mark Brady Ingle, anything to add here?
Will Kalman (and no brush and commutator wear)
Piotr Esden-Tempski @Will Kalman that too, you are right :)
Bhavesh Kakwani @Sophi Kravitz I suppose another important hardware advancement would be higher density batteries
Sophi Kravitz YES! Battery technology has changed over last couple decaddes
Sophi Kravitz Next q is from @Jordan Bunker :What's the most exciting use case you've come across for Paparazzi? Anything unexpected?
Piotr Esden-Tempski that is a good question, I think there were a few massive leaps in the recent history. I think around 2010 the low cost multiaxis mems sensors was one of them: Thank you Apple. :) Others are much more gradual changes that are slowly improving. These would be better microcontrollers, with more computational power in small sizes. Better power electronics (smaller more efficient low Rds on value mosfets) as well as better batteries, but I feel that the batterie field is still improving at a slower rate than the rest of the electronics. Ohh also radio solutions and localization is an interesting field making good progress. But for outdoor and fully autonomous systems we are still bound to the limits of GPS in general.
Rich Burton Here's a good place to see some of the cool uses of Paparazzi UAV systems:
Piotr Esden-Tempski But there are companies trying to push the localization hardware boundries like Swift navigation and their sub centimeter accuracy GPS systems.
Piotr Esden-Tempski but those are not really on the level of consumer affordability
Piotr Esden-Tempski additionally I think we will see more vision based navigation, thanks to the better/smaller cpu's that have enough horsepower to do that on board.
Sophi Kravitz Next question is
Sophi Kravitz How have changing FAA regulations affected making Paparazzi?
Piotr Esden-Tempski Let me answer really quick the exciting things for paparazzi question. I think the support for transitioning VTOL aircraft was and still is quite exiting.
Piotr Esden-Tempski as well as fully outonomous inter aircraft communication
Sophi Kravitz (sorry jumped too fast :)
Piotr Esden-Tempski and much much more that is quite technical and part of the distributed architecture of the system
Rich Burton Apropos to VTOL:
Piotr Esden-Tempski Thanks @Rich Burton exactly. There is also somewhere the TU-Delft vtol they competed with at the outback challenge, it is also an impressive aircraft.
Piotr Esden-Tempski but I think the full autonomy without needing computation or coordination on the ground is the exciting thing
Piotr Esden-Tempski the groundstation ends up being just a monitoring and safety tool
TU-Delft VTOL:
https://blog.paparazziuav.org/2016/09/20/delftacopter-hybrid-drone-on-a-mission/
Piotr Esden-Tempski and the operator ends up being just the conductor of an orchestra that is playing a tune in the sky :)
Piotr Esden-Tempski The FAA issues, that is a hard one to answer. For paparazzi this is not really that relevant as most of the developers are not based in the US. On the other hand I have to say that it is difficult to keep up with the FAA on what is relevant this particular week to be honest. :)
Rich Burton Martin's work in Anartica is pretty amazing:
https://blog.paparazziuav.org/2015/07/17/flying-the-paparazzi-autopilot-blog/
jamesonbeebe Did anyone go to Defcon this year and see "Game of Drones?"
Piotr Esden-Tempski Ohh yes, that is super exciting, the meteorology researchers love that solution.
Piotr Esden-Tempski the advantage of paparazzi is that you can live monitor sensors on board of the aircraft
Piotr Esden-Tempski so the meteorologists can see changes in the temperature and humidity live and adjust the flightplan accordingly
Piotr Esden-Tempski and do not have to fly the aircraft themselves
Piotr Esden-Tempski they just throw it and it starts the mission on it's own
Piotr Esden-Tempski all they do is monitor the meteorological sensors on the groundstation and can decide to survey another air column based on what they see
Piotr Esden-Tempski Martin Müllers aircrafts were being used in the Antarctica for quite a while I think it was the first autonomous micro uav that flew that far north :)
Rich Burton A more direct link to Martin's work: http://blog.pfump.org/?lang=en
Sophi Kravitz So.. back to FAA :)
Piotr Esden-Tempski ohh right :)
Piotr Esden-Tempski I think I will have to mostly pass on that question. Because I was not keeping up with the topic in the recent year or so. We also do not sell fully assembled systems in the US, this means that the adherence with FAA rulings is the responsability of the system integrator and operator. All the autonomous stuff we do is within the hobby rules and common sense. We do not fly in restricted airspace and keep the aircraft under 400ft.
Rich Burton @Piotr Esden-Tempski nice answer. :)
Sophi Kravitz Alright then, a quicker topic: Are there any recommended uav/drone groups in Salem/Portland areas?
Piotr Esden-Tempski Also you can actually keep an eye on an autonomous aircraft which is not the case when you fly FPV that is being considered Drones ^^
Piotr Esden-Tempski I think there is a group of FPV fliers here in eugene and one in portland. But I am not aware of any organized group that would be into the robotic airborne aircraft. I am aware of a few individuals (Rich is one of them) but that is about it.
Rich Burton Here in Portland, any Paparazzi UAV folks tend to gather here:
Sophi Kravitz thanks!
Peter McCloud Awesome, thanks!
Rich Burton ... every other Monday for Dorkbot.
Sophi Kravitz next q is from @Bhavesh Kakwani : What software/algorithms/technology are you using to make the UAVs autonomous?
Piotr Esden-Tempski That is right! :) Come to dorkbot and you will get the biggest concentration of airborne robotics nerds there. :) Rich and I try to go there on a regular basis. :)
jamesonbeebe @Piotr Esden-Tempski Speaking of FPV flyers, could you make a recommendation (privately if desired) on where I can start? I have an idea of a system i'd like to start with, but dont know if it's worth the investment
Sophi Kravitz And related to software/ algorithms/ technology, @johnowhitaker asks Paparazzi looks amazing! I see the UAVs have GPS etc - do they also have sensors to avoid each other when they get very close?
Piotr Esden-Tempski I am not really an FPV person at all. I build robotic airborne aircraft mostly. You can mount a camera and connect goggles to a paparazzi aircraft too, but that is not exactly the goal here. :)
Piotr Esden-Tempski @Sophi Kravitz thanks both of those are quite vast fields of research. :)
Sophi Kravitz can you give a broad answer?
Piotr Esden-Tempski An autonomous aircraft software stack is pretty much an onion of controllers
Rich Burton @Piotr Esden-Tempski maybe discuss the modular architecture of Paparazzi UAV?
Piotr Esden-Tempski and yes modules as rich mentiones
jamesonbeebe lets say I'm a security researcher "Hacker" and want to send a drone out to do some reconissance on a client's wi-fi envirionment
Sophi Kravitz Sheet for discussion here: https://docs.google.com/spreadsheets/d/1r5NrwLhG-H7lLR3-Vn5lqVvnJ3yNiWyhCvSNLfwARoA/edit#gid=0
Piotr Esden-Tempski so first of all you have sensor data, we have tons of drivers that can read many different accelerometers, gyroscopes, magnetometers (compas), barometer (altitude), GPS (position). All that data is being then fed into an estimator. We have at least 4 different estimators that you can choose from including simple complementary filters to extended kalmann filter. The estimator than provides the attitude of the aircraft that the first PID controller operates on. It tries to keep the aircraft level.
Sophi Kravitz nice overview
Bob Miller How do you choose the estimator?
Piotr Esden-Tempski On top of the attitude controller you then wrap around a altitude controller. Around that you wrap then a position or speed controller that then makes the aircraft either stay in place or move to places you want the aircraft to go to.
Piotr Esden-Tempski And on top of that there is the misson block system that is controlling the high level behavioral decisions of the whole aircraft.
Sophi Kravitz how many PIDs?
Piotr Esden-Tempski I think we have three PID controllers at the moment that are involved in a fully autonomous mission. If you have a fixed wing there is even another PID involved in the climb and decent rates for autonomous landing.
Piotr Esden-Tempski so lot's of tuning fun! :D
Sophi Kravitz not autotuning? ;) do they work together?
Frank Buss don't you have a PID autotune feature, as in Cleanflight?
Piotr Esden-Tempski haha yeah, the autotuning ... I think there was some talk about that in the past but we did not get that implemented yet. I think there were other more interesting things to pursue ... on the other hand we do have the INDI controller that only needs one variable to be provided, but that is a bit more complicated topic. :)
Piotr Esden-Tempski So INDI does not need tuning, all it needs to know is what the commanded thrust to a motor results in force applied to an actuator.
Will Kalman What about vortex ring state? Do you detect and counter it? How so?
Piotr Esden-Tempski It is quite an amazing solution that I honestly still can't wrap my head around fully.
Piotr Esden-Tempski This is a reasonably new thing that was added by one of the researchers from TU-Delft. I think there are a bunch of papers that are available on this controller.
Piotr Esden-Tempski Ahh right Ewoud Smeur is the person who added this.
Sophi Kravitz So we have about 10 more minutes for questions, so I'm going to move to the next one
Piotr Esden-Tempski go ahead :)
Sophi Kravitz from @Peter McCloud : For people familiar with the ArduPilot or PX4 eco system, what are the advantages or disadvantges to the Paparazzi eco system?
Piotr Esden-Tempski So the very short answer is: The architecture of being portable, easy to compile and developer friendly was and is the paradigm of Paparazzi. ArduPilot and PX4 are more geared towards an enduser that is not the developer.
Piotr Esden-Tempski Additionally paparazzi architecture, is meant to be a decentralized agent system. The other systems are bit more monolything compared to that.
Peter McCloud Great summary. Thanks!
Rich Burton As someone that adopted Paparazzi UAV in 2010, the most attractive part for me was the software/hardware architecture. It was designed for the start to focus on building autonomous aircraft systems.
Piotr Esden-Tempski Also Paparazzi was started in 2003 and had bit more time to mature and assure it is stable than the other platforms.
Piotr Esden-Tempski It was built by researchers and developers to allow them to itterate quickly and compete at autonomous aircraft competitions.
Sophi Kravitz Next q:
Sophi Kravitz From @msm1089 : Is it possible to get into a career working on drone software using machine learning at the age of 40 without any employed dev experience?
Sophi Kravitz I'm going to jump in here and say everything is possible
Piotr Esden-Tempski yes, same here :)
msm1089 True enough :)
Rich Burton @Sophi Kravitz yes
Sophi Kravitz Last Q from @jamesonbeebe : Any success on remote controlling drones over long distances using DIY Cellular modules?
Sophi Kravitz You are all welcome to stay and chat btw (we encourage it!), but we officially end at a few minutes after the hour
Piotr Esden-Tempski Also at the moment the whole machine learning for UAV is a very hot topic, if you bring something innovative to the table everyone will be very excited about that.
msm1089 Thanks for the encouragement. I have some ideas!
Piotr Esden-Tempski I am not sure about cellular, but a group of Paparazzi developers did fly over some mountintops in the alps in 2005 or 2006 to prove that it is possible.
Rich Burton @Piotr Esden-Tempski searching for the video. Standby .....
jamesonbeebe Never tried it, very feasable... but I'm just getting into drones and making them autonomous by monitoring and gathering data from clients from my "base"
Sophi Kravitz Sorry- I skipped one Q from @Jordan Bunker and one from @Bhavesh Kakwani - but I got one in from each of you earlier in the chat
Piotr Esden-Tempski they had two ground stations on both sides of the mountain, the aircraft took off, disappeared from the GCS on one side of the mountain and after few minutes appeared on the other ... and it all took place fully autonomously :)
Piotr Esden-Tempski Quite an exciting accomplishment.
jamesonbeebe I think i remember that video.... was it done with cellular though? I would think Cell coverage would be pretty spotty at points in mountainous areas
Rich Burton Here's one instance: https://blog.paparazziuav.org/2007/12/27/flying-aircraft-over-internet/
jamesonbeebe thanks @Rich Burton
Piotr Esden-Tempski ohh yes also there is a live demo of flying over the internet: https://www.youtube.com/watch?v=1ei7pkPGkAA
jamesonbeebe I'll check it out
Piotr Esden-Tempski that was part of a talk at the Chaos Communication Congress in 2006
jamesonbeebe very cool
Sophi Kravitz super!
Piotr Esden-Tempski they controlled 3 aircraft from stage that were in thee distinct places in europe
Piotr Esden-Tempski all quite far away from the conference hall :)
Piotr Esden-Tempski one place was in another part of germany and the other in france I think
jamesonbeebe watching now. What about recharging while flying?
Piotr Esden-Tempski there are interesting solutions for that, based on lasers and microwave
Piotr Esden-Tempski but nothing DIY as far as I know
Piotr Esden-Tempski ohh solar is another solution
jamesonbeebe just operating a drone itself takes a power toll and I imagine sending a Video feed and operating a cell modem can really drain the Batteries
Piotr Esden-Tempski there is a guy on youtube building a DIY solar airplane, it is very promising
jamesonbeebe I was thinking Solar.... but is it efficient enough to drop down in a sunny field for a while and start up again when it has enough charge?
Piotr Esden-Tempski usually the electronics are the minor part of the power consumption, propulsion is usually the biggest energy cost
Will Kalman @Piotr Esden-Tempski , do you detect vortex ring state or just ensure to stay away from it in your flight envelope?
Sophi Kravitz I'm going to jump off in a few. Great topic, explained wonderfully. Thank you for hosting @Piotr Esden-Tempski and @Rich Burton !
Piotr Esden-Tempski VTOL might be able to be used in that way, it would have to be quite a large fixed wing to lang on the gears. Multicopters are probably too inefficient to be used in that way. But I am happy to be proven wrong. :)
msm1089 Thank you for hosting Sophi
Sophi Kravitz we'll be posting a transcript shortly :)
Piotr Esden-Tempski @Sophi Kravitz thank you very much for having us! :) Enjoy the Maker Faire! :D
jamesonbeebe Nice chat guys! very informative
matteodl89 thank you!
Piotr Esden-Tempski @Will Kalman no I do not think we do that at the moment.
Discussions
Become a Hackaday.io Member
Create an account to leave a comment. Already have an account? Log In.