MIT Breaks Autonomous Drone Speed Limits By Not Sweating Obstacles

February 12, 2018 by Will Sweatman 29 Comments

How does one go about programming a drone to fly itself through the real world to a location without crashing into something? This is a tough problem, made even tougher if you’re pushing speeds higher and high. But any article with “MIT” implies the problems being engineered are not trivial.

The folks over at MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) have put their considerable skill set to work in tackling this problem. And what they’ve come up with is (not surprisingly) quite clever: they’re embracing uncertainty.

Why Is Autonomous Navigation So Hard?

Suppose we task ourselves with building a robot that can insert a key into the ignition switch of a motor vehicle and start the engine, and could do so in roughly the same time-frame that a human could do — let’s say 10 seconds. It may not be an easy robot to create, but we can all agree that it is very doable. With foreknowledge of the coordinate information of the vehicle’s ignition switch relative to our robotic arm, we can place the key in the switch with 100% accuracy. But what if we wanted our robot to succeed in any car with a standard ignition switch?

Now the location of the ignition switch will vary slightly (and not so slightly) for each model of car. That means we’re going to have to deal with this in real time and develop our coordinate system on the fly. This would not be too much of an issue if we could slow down a little. But keeping the process limited to 10 seconds is extremely difficult, perhaps impossible. At some point, the amount of environment information and computation becomes so large that the task becomes digitally unwieldy.

This problem is analogous to autonomous navigation. The environment is always changing, so we need sensors to constantly monitor the state of the drone and its immediate surroundings. If the obstacles become too great, it creates another problem that lies in computational abilities… there is just too much information to process. The only solution is to slow the drone down. NanoMap is a new modeling method that breaks the artificial speed limit normally imposed with on-the-fly environment mapping.

Continue reading “MIT Breaks Autonomous Drone Speed Limits By Not Sweating Obstacles” →

Delivery Drones Can Learn From Driving And Cycling

February 6, 2018 by Steven Dufresne 10 Comments

Increasingly these days drones are being used for urban surveillance, delivery, and examining architectural structures. To do this autonomously often involves using “map-localize-plan” techniques wherein first, the location is determined on a map using GPS, and then based on that, control commands are produced.

A neural network that does steering and collision prediction can compliment the map-localize-plan techniques. However, the neural network needs to be trained using video taken from actual flying drones. But generating that training video involves many hours of flying drones at street level putting vehicles and pedestrians at risk. To train their DroNet, Researchers from the University of Zurich and the Universidad Politecnica de Madrid have come up with safer sources for that video, video recorded from driving cars and bicycles.

For the drone steering predictions, they used over 70,000 images and corresponding steering angles from the publically available car driving data from Udacity’s Open Source Self-Driving project. For the collision predictions, they mounted a GoPro camera to the handlebars of a bicycle and drove around a city. Video recording began when the bicycle was distant from an object and stopped when very close to the object. In total, they collected 32,000 images.

To use the trained network, images from the drone’s forward-facing camera were fed into the network and the output was a steering angle and a probability of collision, which was turned into a velocity. The drone remained at a constant height above ground, though it did work well from 1.5 meters to 5 meters up. It successfully navigated road lanes and avoided moving pedestrians and bicycles. Intersections did confuse it though, likely due to the open spaces messing with the collision predictions. But we think that shouldn’t be a problem when paired with map-localize-plan techniques as a direction to move through the intersection would be chosen for it using the location on the map.

As you can see in the video below, it not only does a decent job of flying down lanes but it also flies well in a parking garage and a hallway, even though it wasn’t trained for either of these.

Continue reading “Delivery Drones Can Learn From Driving And Cycling” →

Watching The Watchers: Are You The Star Of An Encrypted Drone Video Stream?

January 15, 2018 by Roger Cheng 40 Comments

Small aircraft with streaming video cameras are now widely available, for better or worse. Making eyes in the sky so accessible has resulted in interesting footage that would have been prohibitively expensive to capture a few years ago, but this new creative frontier also has a dark side when used to violate privacy. Those who are covering their tracks by encrypting their video transmission should know researchers at Ben-Gurion University of the Negev demonstrated such protection can be breached.

The BGU team proved that a side-channel analysis can be done against behavior common to video compression algorithms, as certain changes in video input would result in detectable bitrate changes to the output stream. By controlling a target’s visual appearance to trigger these changes, a correlating change in bandwidth consumption would reveal the target’s presence in an encrypted video stream.

Continue reading “Watching The Watchers: Are You The Star Of An Encrypted Drone Video Stream?” →

FCC Fines Drone FPV Maker For Using Radio Spectrum

December 20, 2017 by Al Williams 35 Comments

If you listen to the radio bands in the United States, you might wonder if anyone at the FCC is paying attention, or if they are too busy selling spectrum and regulating the Internet. Apparently however, they are watching some things. The commission just levied a $180,000 fine on a company in Florida for selling audio/visual transmitters that use the ham bands as well as other frequencies.

The FCC charged that Lumenier Holdco LLC (formerly known as FPV Manuals LLC) was marketing uncertified transmitters some of which exceeded the 1-W power limit for ham transmitters used on model craft.

Continue reading “FCC Fines Drone FPV Maker For Using Radio Spectrum” →

Frankendrones: Toy Quads With A Hobby Grade Boost

December 18, 2017 by Tom Nardi 3 Comments

If you’re not involved in the world of remote controlled vehicles, you may not know there’s a difference between “toy” and “hobby” grade hardware. For those in the RC community, a toy is the kind of thing you’ll find at a big box store: cheap, works OK, but lacking in features and build quality. On the other hand, hobby hardware is generally considered to be of higher quality and performance, as well as being more modular. At the risk of oversimplification: if you bought it ready to go from a store it’s probably a toy, and if you built it from parts it would generally be considered hobby grade.

But with the rock bottom prices of toy quadcopters, that line in the sand is having a harder time than ever holding some in the community back. The mashup of toy and hobby grade components is giving rise to the concept of “frankendrones” that combine the low cost of toy hardware with key upgrades from the hobby realm. Quadcopter blogger [garagedrone] has posted a roundup of modifications made to the Bayangtoys X16, a $99 quadcopter which is becoming popular in the scene.

Some of the modifications are easy enough for anyone to do. Swapping out the original propellers for ones meant for the DJI Phantom 3 increases performance and doesn’t even require tools. If you want to go a bit further down the rabbit hole, you can cut off the X16’s battery connector and replace it with a standard XT60. That lets you use standard 3S LiPo batteries, which are cheaper and higher capacity than the proprietary ones the toy shipped with.

If you have a 3D printer, there are also a number of upgraded parts you can print which will bolt right onto the X16. Payload adapters, landing gear, and GoPro mounts are all just a few clicks (and some filament) away. This library of 3D printable parts is made possible in part because the X16’s frame is itself a clone of another toy quadcopter, the popular Syma X8C. So anything listed as compatible with the Syma X8C should work with the X16 (and vice versa).

Finally, if you really want to take the X16 to the next level, you can swap out the flight controller with an open source and better supported hobby grade model. Some of these flight controllers and associated new receivers can end up costing about half as much as the X16 did to begin with, but the vast improvement in performance and capability should more than make up for the cost.

We’ve covered previous efforts to increase the performance of low cost quadcopters in the past, as well as builds that put frugality front and center. It seems that no matter what your budget is a screaming angel of death is available if you want it.

Thanks to [Calvin] for the tip.

Continue reading “Frankendrones: Toy Quads With A Hobby Grade Boost” →

The Zombie Rises Again: Drone Registration Is Back

December 14, 2017 by Jenny List 69 Comments

It’s a trope of horror movies that demonic foes always return. No sooner has the bad guy been dissolved in a withering hail of holy water in the denoeument of the first movie, than some foolish child in a white dress at the start of the next is queuing up to re-animate it with a careless drop of blood or something. If parents in later installments of popular movie franchises would only keep an eye on their darn kids, it would save everybody a whole lot of time!

The relevant passage can be found in section 1092(d) of the National Defense Authorization Act, on page 329 of the mammoth PDF containing the full text, and reads as follows:

(d) RESTORATION OF RULES FOR REGISTRATION AND MARKING OF UNMANNED AIRCRAFT
.—The rules adopted by the Administrator
of the Federal Aviation Administration in the matter of registration
and marking requirements for small unmanned aircraft (FAA-2015-
7396; published on December 16, 2015) that were vacated by the
United States Court of Appeals for the District of Columbia Circuit
in Taylor v. Huerta (No. 15-1495; decided on May 19, 2017) shall
be restored to effect on the date of enactment of this Act.

This appears to reverse the earlier decision of the court, but does not specify whether there has been any modification to the requirements to prevent their being struck down once more by the same angle of attack. In particular, it doesn’t change any of the language in the FAA Modernization Act of 2012, which specifically prevents the Agency from regulating hobby model aircraft, and was the basis of Taylor v. Huerta. Maybe they are just hoping that hobby flyers get fatigued?

We took a look at the registration system before it was struck down, and found its rules to be unusually simple to understand when compared to other aviation rulings, even if it seemed to have little basis in empirical evidence. It bears a resemblance to similar measures in other parts of the world, with its 250 g weight limit for unregistered machines. It will be interesting both from a legal standpoint to see whether any fresh challenges to this zombie law emerge in the courts, and from a technical standpoint to see what advances emerge from Shenzhen as the manufacturers pour all their expertise into a 250 g class of aircraft.

Thanks [ArduinoEnigma] for the tip.

Building A Drone That (Almost) Follows You Home

December 13, 2017 by Lewin Day 9 Comments

There’s a great deal of research happening around the topic of autonomous vehicles of all creeds and colours. [Ryan] decided this was an interesting field, and took on an autonomous drone as his final project at Cornell University.

The main idea was to create a drone that could autonomously follow a target which provided GPS data for the drone to follow. [Ryan] planned to implement this by having a smartphone provide GPS coordinates to the drone over WiFi, allowing the drone to track the user.

As this was a university project, he had to take a very carefully considered approach to the build. Given likely constraints on both money and time, he identified that the crux of the project was to develop the autonomous part of the drone, not the drone itself. Thus, off-the-shelf parts were selected to swiftly put together a drone platform that would serve as a test bed for his autonomous brain.

The write up is in-depth and shares all the gritty details of getting the various subsystems of the drone talking together. He also shares issues that were faced with altitude control – without any sensors to determine altitude, it wasn’t possible to keep the drone at a level height. This unfortunately complicated things and meant that he didn’t get to complete the drone’s following algorithm. Such roadblocks are highly common in time-limited university projects, though their educational value cannot be overstated. Overall, while the project may not have met its final goals, it was obviously an excellent learning experience, and one which has taught him plenty about working with drones and the related electronics.

For another take on autonomous flight, check out this high-speed AI racing drone.