basketclosedownemailfacebookgoogleplushomeinstagramleftlinkedinlocationminusnavphonepinterestplusquoterightsearchstarsubnavticktwitterupyoutube

The Future Commercial Drone: Part 2


In this second part, we shall be looking at the Future Commercial Drone in terms of Safety; Positioning and Metadata; Extra Sensors; and we will also touch on On-Board Processing.

 

Safety

Prop guards must be included within the design of a drone, whether they are easily removable is another issue. Prop guards add an extra bit of safety when bumping into things, or in the unfortunate event of hitting a living being, serious injury can be avoided.

Parachutes should be included as standard. They also need to include the maximum weight of whatever is on the gimbal as they can: stop fatal crashes; save lives; and offer an opportunity to save equipment. The parachutes must be built into the frame as it creates a much safer aircraft.

Transponders have been in use in the aviation industry for years. It is surprising that only a few manufacturers have produced units, and even more astounding is that they are not already required by law. Authorities need an easy-to-access system while in the air or on the ground, which can identify a drone. Once drones are fitted with transponders, rogue drones probably won’t have them so it will be easier to spot their intentions. Or at least it can be pinged by an ATC if it appears on their radar.

Very intense, very high-quality LEDs should be grouped together to flash whichever colours respective CAAs want it to display. Drones need to be able to be identified from the ground and from a significant distance away. Consequently, a large flashing LED cluster can create a very visual signal.

All of the above needs to be built into the frame for the Future Commercial Drone to be safe, even when falling from the sky.

 

Positioning and Metadata

A great deal of time is spent taking GNSS readings to create GCPs (Ground Control Points), especially when large areas are taken into consideration. More modern devices have been developed, such as RTK, PPK, and PPP, and these reduce the need for as many GCP’s. It is essential that these systems be integrated within drone platforms for the Future Commercial Drone.

RTK: Needs a ground station on a GCP with an active transmitter to the drone. Some GCP’s are needed. When a new integer ambiguity resolution procedure occurs, due to a loss of signal, RTK systems do not use external information for basing accuracy estimates.

PPK: Needs a ground station, but less GCP’s are needed which provides cost and time savings. During a flight, when a new integer ambiguity resolution procedure occurs, PPK systems can search from previous and future data relative to that instant.

PPP: No ground station needed. Requires the least amount of GCP’s for any of these systems. However, it is still in experimental phase with a number of Universities and Quest UAV working together to produce a solution.

In essence, these systems will reduce the need for GCPs, time taken on site, and offer the ability to compare data sets to ensure accuracy.

Further issues are created with the Metadata and most systems rely on adding location info in post-production. Why can’t the images be tagged when taken? Usually this is done by hacking the camera, but the solution we need is a dedicated camera with integrated high precision GPS/GNSS data directly from the RTK/PPK/PPP system.

The Future Commercial Drone needs a 50MP camera with the PPP system including relevant data within the images’ metadata.

 

Extra Sensors

Intel and Yuneec’s ultrasonic sensors are substandard and I don’t think they are developing quick enough. The sensors have issues at distance, can’t tell how large an obstacle is, nor whether the drone can bypass it. The drone will also move into other objects due to the lack of sensors offering complete coverage around the drone. With no On-Board Processing, this really is a lame duck system. DJI win again by providing optical sensors which work at a greater distance and are married with onboard processing even in its smaller drone the Phantom 4. Machine Learning should be in every drone which has Extra Sensors.

The Future Commercial Drone needs coverage in 360° on all three axis. With this amount of data captured, the sensors would compliment each other. However, at the same time the processing required is intense and must happen On-Board and in Real-Time for it to be effective.

 

On-Board Processing

Nvidia are nonsense merchants of the highest calibre. “The Drone Goes Where No GPS Can” video is utterly fake and misleading. The pathway the drone takes had been mapped already in different lighting conditions, meaning multiple data sets were available to the drone. Essentially, it’s a waypoint flight using landmarks as points of reference instead of GPS. This drone can only go through one set of woods, and even that needs a lot of work to spoof. All Nvidia proved was that you need a human to map an area before you send a drone. Utterly backward!

Google Tango really isn’t going to achieve the wild fantasies that people have of it. It is a very interesting technology, and at its price point is miles better than everything else. But, again, it needs to be married with high end processors and sensors which create even more problems in terms of weight and heat.

SLAM, is such a headache. The problems inherent with SLAM drones are extensive. But, it is possible to get a single drone to map an area and to process it. It is quite another issue to actually use that data effectively.

The Future Commercial Drone needs excellent On-Board Processing to enable Machine Learning. This is the massive challenge.

 

Conclusion

As we have established, the technology is available, and in use, but not combined. Size is always an issue and consequently, we need miniaturization of all key components for the next drone boom to occur. Smaller components which do exactly the same as the full-size ones. This will allow for more cameras, faster On-Board Processing, Real-Time AI decisions, and above all, a safer aircraft. I believe that we should demand better sensors from Yuneec and Intel or they should give up and stop wasting everyone’s time and use DJI’s or at least include On-Board Processing. Finally, PPK systems must be built into the drone as standard to offer precise positioning.

Why am I writing this article? Because there isn’t a company pushing DJI at all. This is going to be a one-horse race until someone works out how to create a game changer. Here is the template. Build the drone the industry badly needs. Hint! Hint! Bloody Hint!

 

TL:DR for CEOs/CTOs:

  • Airframe: Not important. Just space needed for components and 6 props/motors.
  • Batteries: A minimum of two. Maximum of 4.
  • Flight System: Any reputable three. All through a main controller.
  • Gimbal: Open and deep enough for most cameras. Truly modular.
  • Safety: Parachutes. Transponder. High Quality LEDs. Prop guards. ALL.AS.STANDARD.
  • Positioning and Metadata: PPK with images being tagged with data. Or PPP when it is proven to work consistently. 50mp camera with metadata from Positioning system included in every relevant image.
  • Extra Sensors: 360° coverage in three axis. On-Board Processing. Sensors must be able to share data with each other and offer Machine Learning.