http://www.wired.com/2016/03/dji-phantom-4-smarter-faster-super-sleek/
I don't want to like DJI for the same reason I don't want to like Apple - I like open systems that can be tinkered with and that a community can work on together to improve in the long term. But I'm more and more impressed with what they come up with and how they package it.
With all the recent changes between 3dr and APM, can either or both projects regroup to take them on and provide a compelling price/performance alternative? The solo is great but was too expensive to compete directly against the phantom 3, perhaps this more expensive phantom might actually be a good thing for the solo? Has DJI already won the consumer market for good?
Comments
I don't think for a second the DIY "experiment" is coming to an end.
DJI has obviously sewn up the consumer market. They have the brand name people recognize, and the look of the Phantom has become synonymous with "drone" in the minds of nearly all consumers. The 3DR Solo has been on the market for less than a year, and isn't nearly as well known. It also costs more. So it's at a distinct disadvantage if your definition of "success" equates to "selling ~$1000 drones to consumers". But suppose you want to use a drone for something no one's thought of. Suppose you have specific needs that can only be met through some combination of software changes and specific hardware. Does DJI have something you can use in that case? Not likely. I only mess with drones part time, and I've already encountered situations where I needed to make a drone do something it didn't do out of the box. Both for me, and for people paying me. Being on the APM platform, we were able to make specific changes we needed, and get the drone to do what we needed it to do. I cannot imagine how I would have done any of that using DJI's platform.
I hear it said that digital FPV and collision avoidance should not be too hard but we do not seem close in the Open Source world to having it. As for digital FPV, it seems to take serious effort (and money) to develop this. I have seen the impressive direct wifi efforts here but the bottle-neck is the digital camera link. Even when Auvidea perfect the HDMI converter that works reliably it is still a lot of bits to put together. So most of us are remaining in the analog video world for now. I do not even know how I would even start with obstacle avoidance. As @Fnoop said, many of us will not mess with embedded Flight Controller code but can do things with libraries and companion computers. We just need better tools, clear connectivity standards, and more powerful hardware.
I've really enjoyed following this discussion because of the way it weaves between the conflicting requirements of technology and business. This forum is strongly oriented toward the technology side of the discussion, which is why CA set it up in the first place, but it is fascinating to see how the realization of technological dreams is being played out against the backdrop of commercial reality. As a very long time manager of technology companies, this is my favorite playground.
So FWIW, my understanding of the failure of 3DR and other potentially world beating drone manufacturers to deal head-on with the DJI juggernaut boils down to business execution, not technology. There are numerous examples of superior technology failing when forced to compete against businesses that are better at delivering and servicing what they have. I'm not for one second suggesting that DJI is a world class company when it comes to execution, rather, they are better than the competition, and in a new market space that's all they need to be.
The unfortunate part about this assessment is that changing technology focus to "the next big thing" or "redefining the market" to make the potential look better or coming up with "better products" doesn't actually solve the underlying problem. Instead the real and mundane truth is that you cannot build a company around innovation, it simply doesn't have the right architecture to fight commercial battles. Instead, you FIRST have to build a brilliantly managed, efficiently structured, quality based, world class business machine that lives and breathes customer service. Then you slowly pour innovation into the mix and stir gently.....
Some great commentary here (with the exception of the ever-bitter craid issod). My personal take on this is that 3dr have entered a new market space and have had to pioneer new strategies, technologies, funding and products. It's as much bad luck as anything that they haven't been wildly successful - if a single company - DJI - hadn't already locked up the consumer market they would have been. Personally I don't think they did a good job of targeting the Solo - they've almost deliberately ignored their existing userbase and they badly fumbled the launch, but the product itself remains compelling, pioneering even, and I think it's a very good technical model for the future. Companion computers with integrated wifi/control link and digital fpv, it's a real step forward. I think they made a mistake not releasing the companion computer and controller separately to gain some developer momentum around them, as it is the existing userbase have little interest in the Solo. They probably should have licensed/bought/outsourced the companion computer and gimbal technology rather than trying to do everything inhouse which must have been massively more expensive and time consuming. They took a moonshot and it's not quite worked out yet, but I really hope it does for them. I also don't agree that DJI should just be left to the consumer market, much as I like their hardware. If Google had left Apple and Jobs to the smartphone market they would have a total monopoly by now which would have been catastrophic - worse than Microsoft and Gates - they would (and may well still) force the market into whatever direction plays well for their own profit and gain, not the consumers benefit, and limit technology innovation. People forget that the commoditised, standardised PC market was always massively more successful and cheaper than the proprietary vertical stacked Apple alternative - it was the ipod and iphone that saved Apple, until then they were much of an irrelevance.
3dr have funded the APM community to a huge degree which has been incredibly generous and we've benefited from greatly accelerated development and testing. But it has also to some degree directed where that funding and development has gone to, so now the very generous, soft, gold-plated shackles have been released and the APM community is poorer, but free-er for the future. As some others have said here I think the future is in effectively replicating what the Solo has done very effectively - create a companion computer base that people can program to do all sorts of funky things, and seed it with the basics of digital communication and fpv. Flight controllers are best left to flying, another reason the qualcomm route doesn't make sense. I would love to see a third party like hobbyking or foxtech come up with a next generation digital tx/controller with builtin hires screen - I think the solo missed a trick there. From the hardware side there needs to be some standards, both for the electronics and for the mechanicals. It would be great for the ancient systems like i2c and pwm to disappear and be replaced with CAN or equivalent (quicker than is currently happening), and standardised mounts for gimbals and accessories. Come on, we can't do digital ESCs and motor feedback in this day and age - the FC still has no idea what the engines are doing?! If the hardware community can build snap and replace components that fit a common base, it will be hugely more successful than the current fractured reinvent-the-wheel-each-time model. I'm fed up spending hours upon hours, days upon days putting together the hardware when I just want to fly and code - DIY shouldn't mean having to splice, solder, drill and screw every little component.
Flight Controllers are hard - they involve a deep understanding of electronics, low-level coding, realtime systems, embedded systems, physics, maths etc - there will always be a high barrier to entry and limited set of skilled individuals able to work at this level. However, if a lot of the higher-level functionality can be broken out to the companion computer (as I believe the Solo does, for example), then there is a far lower barrier to entry an a lot more people can contribute. Personally I have little hope of making meaningful contribution to the current APM codebase, but hopefully I can be much more helpful at the companion level, and I suspect this is the case for the vast majority of DIYers here. The ingredients are all around us - cheap SBCs like the raspberry, dronekit, and the ever increasing amount of available and cheap accessories and peripherals. Then there are the incredibly powerful (and relatively cheap) next-gen SBCs like the TX1 and snickerdoodle which will open up completely new horizons of what is possible. We just need to put together the recipes. Compared to the magic that happens up in the FC, using the companion computer to do crude SLAM or object avoidance should be relatively simple, digital FPV with OSD should be simple and straight forward, especially when using boards like the TX1 which already have mature libraries for this kind of thing. What it needs is the community to start collaborating and putting together the building blocks and then work on them in an evolutionary and iterative manner.
My 2c :)
@Rob, Yes, I am not clearly mentioned what I want to response.
What do you think on next feature for ArduCopter development?
I do not think that is not kind of throw mode. I like throw mode for surprising somebody;-)
My thought is something like this,
1) Real support of flow sensor for non GPS environment.
2) RTK type precision GPS positioning.
3) Obstacle avoidance, (machine vision or ultra sound)
4) Accurate altitude.
@Jiro, I'm not sure what you're saying. You don't care about the processor board, but then you complain you can't see what the next platform is?
Well 3DR obviously saw this coming and that is why the are using SnapDragon Flight. I am not really familiar with @Jerry Giant's suggestions for what is needed (need to do some reading on it) but it seems that some aspects of this are being kicked up the food chain away from the kind of embedded development we have seen thus far. Does anyone else have an idea as to what hardware targets and systems will accomplish obstacle avoidance for Ardupilot?
@Rob
Yes, you said that before. Actually, we do not need to care about selection of processor board. The problem is that we can not see what is going on the ArduPilot and what is the next platform to develop on ArduPilot. May be nobody knows about it.
By the way, another one off-topic : who is developing on Hololens/whatever a gimbal system reacting to your head orientation?
Snapdragon Flight is not the brightest hardware option for the future of ArduCopter because of Qualcomm's stance on copyright. We can't even distribute compiled binaries.
-
1
-
2
-
3
-
4
of 4 Next