The future of the Micro Drone

I see a very exciting future for the micro drone! Say the size and weight of this Hobby King Q-Bot quadcopter. The reason being these small drones are far quieter and offer far less danger to people, aircraft, cars, motorcycles, children and even pets!

I see a world of useful and safe personal drones with all the features of arducopter and then some that everyone can enjoy safely! I see a lot of fun being had with micro drones WITHOUT any regulations needed.

I'm willing to bet this is the true future of the drone revolution! The micro drone!

Sure I see what a lot of people want to do commercially with larger drones, but the true future of the personal drone must be the micro drone!

Have you ever flown a Q-Bot? The flight times are actually great! They are a heck of a lot of fun too. All I'm saying really is I'd like to see more focus on the tiny drone with no more weight than the Q-Bot. Sure I see a lot of small drones but anything that weighs more than a Q-Bot is going to be just a little bit too much momentum.

The future of the personal drone is without a doubt the micro drone and nearly everyone will have one of these.

Just pull the back off your cellphone and throw it in the air it's a micro drone!

Hey Guys!

Looks like there is some interest here for a micro-uav controller! I'd be willing to start a group if we can get at least 3-4 people to dedicate some time to it to make it progress. I'd also be happy to put together the hardware prototypes, provided we have some dedicated software programmers! I have all the SMD gear for that here apart from a pick n place, but can outsource it easily once we need numbers. I can come up with the PCB design but would want/need help on testing/optimising etc.

But first we'd need to do a bit of a hardware concept. :-)

FPGAs or DSP could offer a significant boost in capability and performance, but they'd have to be programmed nearly from scratch, and we can't simply port APM code. There's only a few guys I know have even attempted this, even less that have been successful. I'd say to make FPGAs work we'd need some 5-10 guys full time...but I'm not a coder so I could be wrong. I just know that coding and debugging is a time sink and without the right people the project will go nowhere fast.

I think the far easier and faster approach would be to leverage APM code, which means we're either limited to running it on Nuttx based ARM processors, Like STM32 etc, or Linux based SBCs like the Raspi or Odroid etc. either of which are way to physically big and heavy for this application. What we need is more along the line of a Gumstix/Lisa/s size board with flight control, but cheaper and with camera.

For small size we need to keep as much of the circuitry in silicon and off the PCB. So we need SOC (System on Chip) solutions for this. So my ideas thus far are:

There is one very new CPU/MCU I know of atm that could be the contender for this called the iMX 7 from freescale that has both a Cortex M4 MCU (like the PXH) for real-time AP functions and in the same die a dual core 1Ghz Cortex A7 CPU. This way we could run APM both on RTOS on the M4 MCU (Nuttx runs on Freescale M4's) and in Linux on the A7 CPU on the same hardware! It could also incorporate failover etc if desired.

Camera interfacing is also onboard along with all processors IO routed out. Apparently it also uses 30% less energy as the same ARM's from other manufactures and has built in power management which allows for dynamic running of cores as required (so linux cores can sleep while the MCU runs things). The modules are available in 19x19mm or 12x12mm packages and have MIPI camera, and MPD display outputs (but no HDMI) support. They also have eMMC 5.0 and DDR3 memory controllers and support either Linux or Android OS on the CPU. So both Linux and Android developers could profit, maybe even run both at the same time on separate CPU's on the dual core if it makes sense to do so for the desired app. I think the dual core will be sufficient for most apps, and if not adding a secondary CPU onboard or only using it to compress information to be sent to the ground should be possible and keep it future proof. Imaging is the bandwidth/processor hog.

L138-FX-225-RC Critical Link LLC | L138-FX-225-RC-ND DigiKey Electronics

For the PCB layout I was thinking of using a single notebook formfactor SO-DIMM DDR socket (68x30mm) to allow daughter boards and connections to made to the PCB (as above). So all IO, apart from the camera MIPI/GPS/SD would pass through the SO-DIMM socket (to keep price and complexity down) and pending on application daughter boards could already include ESCs, BEC, Power monitorng etc which means a quad setup could consist of just the CPUboard connected to a daughterboard (say 6mm thick overall) with a single battery connection and 4/6/8 3Phase power leads coming out of it for motors. The camera would be attached to the mainboard MIPI connector directly as well as a serial port for GPS (both being RF sensitive) and have both EMMC and uSD card connectors for memory expansion. External GPS allows for RTK compatibility and mounting flexibility.

I'd try to populate only one side of the SO-DIMM CPUboard (the side facing the daughter board) leaving the opposite side available to attach HopeRF, WIFI, 3G, GPS modules on pads as desired (Even with hand soldering). This also allows for RF connection access (uRF) and RF cooling. CPU Cooling will need to be considered but I'd expect will be less than RF. The sensors could be external and dampened, as well as on the mainboard, but given the small overall size it's easier to dampen the whole assembly and attach it to the battery mass by isolating the motor arms instead.

For the camera we'd just run a rolling shutter CMOS, maybe the 13MP OV13850 ($5) with 4K output. (although we'd need to confirm if the i.MX can process 4K at all, but for FHD it should be fine) I'd prefer taking a higher res imager to allow for EIS (electronic image stabilisation) algorithms to get rid (most) of the need for a stabilising gimbal, and allow for better still imaging as required for aerial imaging. 4k Could then be an upgrade further along, which could be as easy as clipping in a new hardware SO-DIMM mainboard. A simple compact 1 axis tilt for the camera module would be nice so it could be used for FPV, aerial imaging, optical flow based navigation etc and be immune to vehicle pitch changes. A secondary camera is also possible via the SO-DIMM header, via USB/LAN etc.

For comms I'd concentrate on high power wifi for a micro-UAV. It is internationally accepted and will offer ranges easily over 1km in LOS ops. HopeRF can be optional as can GSM/Sat/4G etc. Wifi also has the bandwidth and latency for FPV now (if done right) plus is directly compatible with wifi ground devices like tabs/phones etc. A ground "daughterboard" could be used for custom RC controllers or as base stations for interconnectivity to camera/pilot controllers etc. Daughterboards could also have more than one SO-DIMM socket to accommodate more that one CPUboard or allow addon comms/FPGA/SDR/Sensor/FLIR hardware.

Device can obviously also support industrial or security appilcations e.g. OpenCV or security camera with direct control of automation, or action/car cam with sensor logging etc. Price target for hardware would have to be under $100, I'd like to aim for $50-80 sale plus camera and daughterboards. Initially it will be a no profit marketing model but needs to cover cost, just to get the hardware out there to make it the dominant micro drone hardware. Total RTF platform say around $300. Keeping hardware cost down also pacifies potential cloners ;-)

So if we can get the interest (if anyone is still reading this!) then put your're names down here and I'll start a new group and thread to discuss further. BTW it will be open source under a creative commons as the only way this will work is if there enough people contributing, with a possible Kickstarter/Indeigogo hardware startup. We'll need a name too.

Regards JB

You need to be a member of diydrones to add comments!

Join diydrones

Comments

JB September 17, 2015 at 11:39pm

Hey Guys!

Looks like there is some interest here for a micro-uav controller! I'd be willing to start a group if we can get at least 3-4 people to dedicate some time to it to make it progress. I'd also be happy to put together the hardware prototypes, provided we have some dedicated software programmers! I have all the SMD gear for that here apart from a pick n place, but can outsource it easily once we need numbers. I can come up with the PCB design but would want/need help on testing/optimising etc.

But first we'd need to do a bit of a hardware concept. :-)

FPGAs or DSP could offer a significant boost in capability and performance, but they'd have to be programmed nearly from scratch, and we can't simply port APM code. There's only a few guys I know have even attempted this, even less that have been successful. I'd say to make FPGAs work we'd need some 5-10 guys full time...but I'm not a coder so I could be wrong. I just know that coding and debugging is a time sink and without the right people the project will go nowhere fast.

I think the far easier and faster approach would be to leverage APM code, which means we're either limited to running it on Nuttx based ARM processors, Like STM32 etc, or Linux based SBCs like the Raspi or Odroid etc. either of which are way to physically big and heavy for this application. What we need is more along the line of a Gumstix/Lisa/s size board with flight control, but cheaper and with camera.

For small size we need to keep as much of the circuitry in silicon and off the PCB. So we need SOC (System on Chip) solutions for this. So my ideas thus far are:

There is one very new CPU/MCU I know of atm that could be the contender for this called the iMX 7 from freescale that has both a Cortex M4 MCU (like the PXH) for real-time AP functions and in the same die a dual core 1Ghz Cortex A7 CPU. This way we could run APM both on RTOS on the M4 MCU (Nuttx runs on Freescale M4's) and in Linux on the A7 CPU on the same hardware! It could also incorporate failover etc if desired.

Camera interfacing is also onboard along with all processors IO routed out. Apparently it also uses 30% less energy as the same ARM's from other manufactures and has built in power management which allows for dynamic running of cores as required (so linux cores can sleep while the MCU runs things). The modules are available in 19x19mm or 12x12mm packages and have MIPI camera, and MPD display outputs (but no HDMI) support. They also have eMMC 5.0 and DDR3 memory controllers and support either Linux or Android OS on the CPU. So both Linux and Android developers could profit, maybe even run both at the same time on separate CPU's on the dual core if it makes sense to do so for the desired app. I think the dual core will be sufficient for most apps, and if not adding a secondary CPU onboard or only using it to compress information to be sent to the ground should be possible and keep it future proof. Imaging is the bandwidth/processor hog.

For the PCB layout I was thinking of using a single notebook formfactor SO-DIMM DDR socket (68x30mm) to allow daughter boards and connections to made to the PCB (as above). So all IO, apart from the camera MIPI/GPS/SD would pass through the SO-DIMM socket (to keep price and complexity down) and pending on application daughter boards could already include ESCs, BEC, Power monitorng etc which means a quad setup could consist of just the CPUboard connected to a daughterboard (say 6mm thick overall) with a single battery connection and 4/6/8 3Phase power leads coming out of it for motors. The camera would be attached to the mainboard MIPI connector directly as well as a serial port for GPS (both being RF sensitive) and have both EMMC and uSD card connectors for memory expansion. External GPS allows for RTK compatibility and mounting flexibility.

I'd try to populate only one side of the SO-DIMM CPUboard (the side facing the daughter board) leaving the opposite side available to attach HopeRF, WIFI, 3G, GPS modules on pads as desired (Even with hand soldering). This also allows for RF connection access (uRF) and RF cooling. CPU Cooling will need to be considered but I'd expect will be less than RF. The sensors could be external and dampened, as well as on the mainboard, but given the small overall size it's easier to dampen the whole assembly and attach it to the battery mass by isolating the motor arms instead.

For the camera we'd just run a rolling shutter CMOS, maybe the 13MP OV13850 ($5) with 4K output. (although we'd need to confirm if the i.MX can process 4K at all, but for FHD it should be fine) I'd prefer taking a higher res imager to allow for EIS (electronic image stabilisation) algorithms to get rid (most) of the need for a stabilising gimbal, and allow for better still imaging as required for aerial imaging. 4k Could then be an upgrade further along, which could be as easy as clipping in a new hardware SO-DIMM mainboard. A simple compact 1 axis tilt for the camera module would be nice so it could be used for FPV, aerial imaging, optical flow based navigation etc and be immune to vehicle pitch changes. A secondary camera is also possible via the SO-DIMM header, via USB/LAN etc.

For comms I'd concentrate on high power wifi for a micro-UAV. It is internationally accepted and will offer ranges easily over 1km in LOS ops. HopeRF can be optional as can GSM/Sat/4G etc. Wifi also has the bandwidth and latency for FPV now (if done right) plus is directly compatible with wifi ground devices like tabs/phones etc. A ground "daughterboard" could be used for custom RC controllers or as base stations for interconnectivity to camera/pilot controllers etc. Daughterboards could also have more than one SO-DIMM socket to accommodate more that one CPUboard or allow addon comms/FPGA/SDR/Sensor/FLIR hardware.

Device can obviously also support industrial or security appilcations e.g. OpenCV or security camera with direct control of automation, or action/car cam with sensor logging etc. Price target for hardware would have to be under $100, I'd like to aim for $50-80 sale plus camera and daughterboards. Initially it will be a no profit marketing model but needs to cover cost, just to get the hardware out there to make it the dominant micro drone hardware. Total RTF platform say around $300. Keeping hardware cost down also pacifies potential cloners ;-)

So if we can get the interest (if anyone is still reading this!) then put your're names down here and I'll start a new group and thread to discuss further. BTW it will be open source under a creative commons as the only way this will work is if there enough people contributing, with a possible Kickstarter/Indeigogo hardware startup. We'll need a name too.

Regards JB
Developer

Randy September 17, 2015 at 6:16pm

If we're interested in getting Copter running on it (which would be fun I think), a port is very easy if it's running Linux. The next easiest is if it's got the same CPU as the Pixhawk's main FMU CPU or the Pixhawk's I/O CPU so that we can run NuttX on it (which is what the Pixhawk's use). Sorry I don't immediately know the CPU names.
HeliStorm September 17, 2015 at 6:05pm

I have seen that RotorX racer. Pretty amazing piece of kit. Next step would be for the likes of a Mobius or other small sport cam manufacturer to make a virtual gimbal camera.
Daniel Worger September 17, 2015 at 5:58pm

Have you guys seen the micros already coming out of the FPV drone racing community? RotorX have developed a truely powerful brushless motor/prop combo that is revolutionizing what you can do in this form factor.

http://www.rcgroups.com/forums/showthread.php?t=2391308

I have no issue carrying a Mobius or Runcam for 1080p video. https://www.youtube.com/channel/UCnE2USiJqaojv1S42l5gSdQ

RotorX will also be coming out with a ready to fly version this Christmas.
Monroe Lee King Jr September 17, 2015 at 5:06pm

I can't stress how much fun the Q-Bot is. It has an acro mode and even does flips and rolls quite well. It's just too much fun if there is a better flying micro drone I don't know what it is and I like using my 9x with it to trim it out so it fly's real nice.

Yes, someone start a micro drone group and I will put some time into it. I've started too many groups here someone else do it this time? Just link to this thread and I'll be happy.

But yes! I think we can actually do this. Let's take a shot at a DIY Drones MICRO!

I think it might be possible to get some 3DR backing if we do well. Bet cha :)
Jiro Hattori September 17, 2015 at 3:45pm

I want give a try some of those micro copter. However, Q-Bot is back order at HK.

Is there any recommendation for this kind of micro?
HeliStorm September 17, 2015 at 11:00am

Maybe we could start a group in here to further discuss? Just a thought.
Gary McCray September 17, 2015 at 10:17am

Hi Fnoop,

There are a lot of tiny drones that are badly affected by wind, but that is not because they are small.

It is because they have a large wind profile relative to their mass.

There are also small drones for which this is not true. The Blade 200QX among them.

It is remakably stable in winds and competitive with any other copter.

It is heavy for a small quadcopter and it has a very aerodynamically designed frame.

This and a responsive flight controller are all we need to solve the wind problem.

Like I said top down design.

And Monroe, can't think of anyplace better equipped to do it than right here.

It will help of course if we can get the truly serious devs to share our view, I do know some of them are starting to come around.

Best Regards.

Gary
earthpatrol September 17, 2015 at 10:04am

@Jiro_Hattori these tiny, fully autonomous capable flight controllers have been around for a few years now. Check them out. They work really well.

Lisa/S Flight Controller
Hans H. September 17, 2015 at 9:50am

indoor GPS? centimetre accurate position? Link to chip? You know, if you add a light saber from star-wars you can also trim the hedges around the house. :D

of 5

This reply was deleted.