Blogs

All Posts (14054)

Sort by
3D Robotics

Stanford's autonomous helicopters teach themselves how to fly

Posted by Chris Anderson on September 1, 2008 at 8:21pm

Stanford computer scientists have developed an artificial intelligence system that enables robotic helicopters to teach themselves to fly difficult stunts by watching other helicopters perform the same maneuvers. The result is an autonomous helicopter than can perform a complete airshow of complex tricks on its own. Much more in the full article here.
Read more…
3D Robotics

Easy datalogging with Arduino

Posted by Chris Anderson on September 1, 2008 at 10:00am
Doing traditional datalogging with Arduino is a pain. It only has 1k of RAM, and half of that is used by the program, so you can't store any significant data in its own memory. Meanwhile, writing to a SD card, as the Adafruit GPS logger does, takes pretty much all of its computational power, leaving little left for useful work. This is not unique to Arduino--it's true for any 8-bit embedded processor. You can add I2C memory boards, but I've never tried that (there's no code available at that link) and you'll still have to find a way to get the data back out of those chips and into your PC. I'm writing a datalogger now to do altitude comparisons of GPS, relative pressure sensors and absolute pressure sensors, all head-to-head. That means the Arduino has to read and parse GPS, read analog and read and parse SPI, along with writing to SD card memory. Which is a lot. Wish me luck. But the easy way is just to do this while you're on the ground (or in my case, in a car and driving up and down hills) and the Arduino is connected to your laptop is to just print all the data to the IDE's serial console. By default it only records 500 lines, but you can go into the preferences file (you can see its location from the Preferences menu item) and change "console.length=500" to whatever amount of data you want to record. When you're done you can just copy the text from the console and paste it into a spreadsheet to analyze. You can deal with SD cards and inflight recording some other day....
Read more…
3D Robotics

Quickie review of the Pololu Arduino-compatible wheeled bot

Posted by Chris Anderson on August 30, 2008 at 2:00pm
I just got the new Pololu 3pi wheeled bot that I posted about last week, and it's every bit as cool as I'd hoped. It's quite small (those are AAA batteries in the picture, not AAs). It's basically one beautifully designed round circuit board that holds all the components and also sports lots of available holes, some of which are connected to I/O pins for adding your own sensors. The instructions and example programs are also excellent. It's really fun to make a maze and watch it solve it, and the bot is tremendous at line following. The obvious thing to add to it is a "proto shield", or expansion board with a breadboards, and I'm pleased to see that Pololu is developing one. Highly recommended, and at this price ($99) it's well worth getting one.
Read more…
3D Robotics

First bot-on-bot attack in Iraq

Posted by Chris Anderson on August 27, 2008 at 9:00pm
We don't cover military UAVs much here, but this one is too notable to overlook (from Gizmodo): "For the first time in history, an unmanned machine has engaged and destroyed another unmanned machine in real combat. It sounds like science fiction, but it happened a week ago in Iraq, when a MQ-9 Reaper killed a remote controlled vehicle carrying a bomb. The MQ-9—operated by the 46th Expeditionary Reconnaissance and Attack Squadron—destroyed the RC car using a laser-guided 500-pound GBU-12 bomb in southeast Iraq. The commander of the squadron, Lt. Col. Micah Morgan, said that they "searched for, found, fixed, targeted and destroyed a [threat] with just one aircraft." With an entire wing of UAVs in the US Air Force, we won't be surprised to see full-scale unmanned and robotic battles in the near future. The question is: if nobody is going to die—except probably civilians on the ground—what's the bloody point of real war? Why not just organize a videogame league and get over with it? [Aviation Week]" [UPDATE: Actually, on reading the AvWeek story, I think Gizmodo may have got a bit overheated. It doesn't say that the ground vehicle was unmnanned, only that the explosive devices was remotely operated. Although you can't entirely blame Gizmodo, since the AvWeek headline is "Unmanned-On-Unmanned Combat Emerging In Iraq"]
Read more…
3D Robotics

FAA rejects petition to allow small UAVs for natural resources management

Posted by Chris Anderson on August 27, 2008 at 3:47pm
Here's an update to the PineCreek petition to the FAA to allow small UAVs for natural resources management. It was denied. Excerpt from the letter follows: Mr. Robert Blair CEO PineCreek Precision 15401 Blair Hollow Lane Kendrick, ID 83537 Dear Mr. Blair: In your May 2, 2008, letter you petitioned the Federal Aviation Administration (FAA) to amend part 91 of Title 14, Code of Federal Regulations (14 CFR) to allow the use of small Unmanned Aircraft Systems (UASs) for Natural Resource Management. The FAA already has an activity underway that we believe will address your concerns. The FAA has established an Aviation Rulemaking Committee (ARC) for small UASs. The ARC is tasked to develop a proposal for the use and operation of small UASs in the National Airspace System. We anticipate the ARC’s recommendations within a year, and will consider those recommendations as the basis for potential future rulemaking activity. The committee will define generic requirements for the operation, maintenance, and certification of small UASs, along with certification of the crewmembers operating and maintaining them. For the reason above, we hereby deny your petition for rulemaking in accordance with 14 CFR § 11.73. Accordingly, Docket No. FAA-2008-0717 is being closed.
Read more…
3D Robotics

A new (and inexpensive) Arduino-compatible wheeled bot

Posted by Chris Anderson on August 27, 2008 at 3:09pm

We're all about the flying things here, but since Arduino robotics is also a focus I thought I'd pass this along. Pololu has just released a nifty two-wheeled bot based on the Atmega168. It's under $100 and looks really fun. Here's the description from the site: "The Pololu 3pi robot is a complete, high-performance mobile platform featuring two micro metal gearmotors, five reflectance sensors, an 8×2 character LCD, a buzzer, and three user pushbuttons, all connected to a C-programmable ATmega168 microcontroller. Capable of speeds exceeding 3 feet per second, 3pi is a great first robot for ambitious beginners and a perfect second robot for those looking to move up from non-programmable or slower beginner robots." Here are some annotated shots of the board:

And a demonstration video:
Read more…
3D Robotics

New Draganfly semi-autonomous heli-thing looks awesome

Posted by Chris Anderson on August 26, 2008 at 9:30pm

It's got six rotors on three arms, 11 sensors (including GPS) and flies like a dream. God knows what it costs, but if you have to ask you can't afford it. [UPDATE: I just got an email from them. They say "The base model starts at $14,995.00."] The new Draganflyer X6 is designed to carry cameras to get smooth aerial video or still photography, anytime, anywhere. The GPS allows you to fly it with a single stick to simply go up and down in a vertical column. Obviously it could fly autonomously to waypoints, but once again that feature is not enabled (presumably to avoid legal/regulatory issues). Loads of videos here. It even has a custom controller with a screen that shows inflight telemetry and other data:

Read more…
3D Robotics

Finally, a decent (albeit expensive) competitor to the Eagle PCB software

Posted by Chris Anderson on August 26, 2008 at 4:30pm

I've just been playing with the demo version of Pulsonix, which is the best competitor to the aging Cadsoft Eagle PCB design software that is the standard in our industry. It looks fantastic, improving on Eagle in loads of ways such as easy viewing of the front or back side of the board without clicking on dozens of check boxes (a pet peeve of mine about Eagle) and built-in 3D viewing. It can also import Eagle files and part libraries with the use of a translation utility. (A full list of features is here) The only downside (and this is a big one) is that the free demo has save disabled, and the cheapest full-featured version of the software costs $475. [CORRECTION: This price actually refers to a lower-end version of the software, called Easy PC. Pulsonix is the top end of their range and sells for around $2,000. The bottom end of the range is the free, stripped-down but fully functional version of this software, which is called PCB Artist. It's very good, but doesn't support the Eagle file and library conversion utilities.] Here's the description from Pulsonix's US distributor, Advanced Circuits: …the 1st completely NEW, High Level combined Schematics Capture & PCB Layout Software announced in many years. • Pulsonix has been developed from the ground up by PCB design industry professionals using the very latest techniques in graphics and data handling. • Competitive products originate from the 80’s and 90’s and look antiquated by comparison. • Pulsonix uses Microsoft Office menu structure and is immediately familiar and attractive. • Usually does not require formal training. Designed for the casual user and the professional. • Directly imports schematics, PCB Designs, and Libraries from other PCB software products. • Built in features at no extra cost; ie, Design Variants, DRC, and Report Maker. • Can be purchased with or without an Autorouter. • Product includes all features for design, checking and manufacturing PCBs. All outputs are included (Gerber, ODB++, Excellon, etc.) • Pulsonix is delivered with an informative Users Guide and up to date, context sensitive on-line HTML help.
Read more…
3D Robotics

New RangeVideo OSD + return-to-launch autopilot for $299

Posted by Chris Anderson on August 26, 2008 at 11:00am

This week Range Video is expected to release a cool new combo: an On-Screen Display (OSD) board with a built-in return-to-launch autopilot, failsafe and and visual guides to GPS waypoints. The product page is here and a long discussion thread about it is here. It has a temperature sensor and current sensor to monitor battery and ESC status. Note: this OSD board does no stabilization (no gyros, accelerometers or thermopiles), so you either need an inherently stable aircraft like an EasyStar or to add a FMA CoPilot to handle the stabilization function. But it does have a 5hz GPS, so that should work for very docile aircraft. Also it is not an programmable autopilot. Although you can enter waypoints, it will simply give you directional indicators on the OSD, not fly the plane to those waypoints itself. The only waypoint it will fly to automatically is the launch position. Here's the promotional video:
Here's the writeup from the site: $299 purchase includes: (1) RVOSD with GPS (1) Current sensor with 150 mm wire (1) Temperature sensor (1) IR remote (controls OSD menu) (3) 150mm male to male servo wires ((to connect OSD to Rx) (3) 300mm male to male servo wires (to connect OSD to Rx) (1) KX131/KX191 camera to OSD connector (1) KX171 camera to OSD connector (1) 300mm male to bare lead servo cable (for making transmitter to OSD wiring harness) Introduction: The RangeVideo on-screen display (abbreviated OSD) is a small ( 8 x 4 x 3 cm) device which overlays GPS telemetry and and other relevant data onto a live video.This device is a must have for FPV flying! ----------------------------------------------- Features: 5Hz GPS Autopilot Failsafe Navigation and battery telemetry. Multiple screens Integrated power supply and filter New features added (update Aug 4, 2008): 1. Amateur radio call sign display. 2. Way point navigation (heading and distance to waypoint) 3. Shadows added to the display
Read more…
3D Robotics

Microsoft selling Virtual Earth aerial imagery for stand-alone use

Posted by Chris Anderson on August 25, 2008 at 11:06am
Microsoft just announced that you can now buy the high-resolution aerial photography that it uses in Virtual Earth, without having to license the whole Virtual Earth package. I doubt our community would be in the market for this, but it does speak to the demand for such imagery, which our community could provide. The pricing was not disclosed (apparently it varies from place to place and the age and quality of the imagery). Does anyone know what the range is?
Read more…

Stanford Autonomous Helicopter Project

Posted by Howard Gordon on August 24, 2008 at 7:30am
tempest-thumb.pngFrom Yann's Techno Toy Blog ...Stanford Autonomous Helicopter Project has developed new reinforcement learning techniques to stabilize and control their helicopters. The developers have been able to teach their heli to perform just about every aerobatic figure imaginable. However, the heli has no on-board intelligence. Instead, data is fed from onboard GPS and IMU with serial outputs directly into Xbee Pro modules. The signals are received by a PC on the ground, which enables operator control via the training port of a conventional R/C transmitter.
Read more…

XBee price cut

Posted by Jack Crossfire on August 24, 2008 at 4:00am
The 2.4Ghz modules may be heading up to $36, but the cost of 900Mhz access will be going down in 2009. All the way down to $39 for a 900Mhz module which is upgraded to 115200 baud but range reduced to 1.8 miles.http://www.digi.com/products/wireless/zigbee-mesh/xbee-digimesh-900.jsphttp://www.digi.com/products/wireless/point-multipoint/xbee-pro-900.jspIt will be 5V tolerant on the UART pins, but you better keep those voltage dividers around until 2009. The $32 XBees are not 5V tolerant & quickly burned out if U followed the social network hits on Google that said they were.
Read more…
3D Robotics

New home pages for ArduPilot, ArduPilot Pro and BlimpDuino

Posted by Chris Anderson on August 23, 2008 at 6:45pm
As we get closer to releasing products, I've set up home pages for ArduPilot, ArduPilot Pro and BlimpDuino. They're created as big posts at the moment, but they'll probably migrate to proper minisites with wikis once the products are out. They're all linked to from the front page of this site. (I've also retired GeoCrawler 4, which was based on the PicoPilot, in anticipation of leaving all the UNAV autopilots for AttoPilot as soon as it's released.)
Read more…

Paparazzi installed in airframe, nearly ready to fly.

Posted by Cory on August 20, 2008 at 11:40am
I am happy to report that I have installed the paparazzi autopilot in one of my airframes and have done some ground testing. So far the results are positive. The telemetry works well (at short range at least), and the aircraft reports proper position and attitude. When the aircraft is held at an angle, the control surfaces move to compensate, and the behavior of the control surfaces looks correct.In the picture below, all of the autopilot system except for the radio modem can be seen.

In this view of the fuselage interior, the radio modem is visble.

This is a detail shot of the installation of the horizontal infrared sensor.

This is a detail shot of the installed vertical IR sensor.

And finally, a shot of the interior of an RC reciever that has been hacked to output the PPM signal on the channel 8 PWM pin. The cut trace to channel 8 is hidden by the capacitor.

I have a bit more ground testing and configuration to complete, and this plane will be ready to fly.
Read more…
3D Robotics

New version of the ArduPilot board

Posted by Chris Anderson on August 19, 2008 at 9:00pm
Jordi and I are preparing the final production version of the ArduPilot board, which involves the usual debugging and last-minute feature adding. The problems we've found are largely in power and noise management--we've needed to isolate the servos from the processor parts, so high current draws don't introduce voltage dips. You'll notice a couple extra diodes and capacitors on the board to handle that. As for features, here's what's new:
  • All four RC-in channels (plus the autopilot on/off channel) can now be read by the CPU. This allows the autopilot to read the initial throttle and rudder position when you switch to autonomous mode, so it can maintain speed and heading. The other two channels can be used for whatever you want, such as triggering a camera sequence or dropping an object. (Remember that the aileron and elevator channels are seperately controlled by the FMA Co-Pilot)
  • There are now status LEDs for the failsafe (on/off) and the GPS (satellite lock).
  • There is now a trim pot on the board so you can adjust the autopilot on/off position to suit your particular RC system.
Here's the schematic, in Eagle 5 format. I'm not including the PCB file yet because we're working with our commercial manufacturing/retail partner on that, and I don't want to reveal those details until we're ready to release. But rest assured that we'll make both the schematic and PCB files available here when the final product is available. For those who don't have Eagle, here are the schematics in png form (they're shown as two pages, but it's pretty easy to see which wires connect across the pages):

Read more…
3D Robotics

Learn about LabView for UAVs at AUVSI meeting

Posted by Chris Anderson on August 18, 2008 at 6:51pm
Jordi and I are planning to use LabView to create the open source ground station software for ArduPilot (National Instruments has kindly donated a professional edition licence to DIY Drones). We saw a lot of UAV teams, especially academic and research groups, using LabView at the AUVSI annual meeting in San Diego, and we're starting to see more interest in the open source side of it, including the OpenG community. For those in the San Diego area who are interested in learning more about LabView for UAVs, you may be interested in this upcoming AUVSI local chapter meeting: San Diego Chapter Meeting May 21, 2008 11:30 am -1:00 pm Southwestern Yacht Club, Point Loma, California Program: National Instruments software and I/O tools for autonomous platforms and payloads. Brief overview of National Instruments Case studies: Virginia Tech/TORC 2007 DARPA Urban Challenge Vehicle, Nexans Spider underwater dredging vehicle, Software and I/O tools overview, and Additional Application Examples Presented by: Charlie Knapp, US Unmanned Systems BDM, National Instruments Please visit our website for pre-registration.
Read more…
3D Robotics

ArduPilot code now in alpha

Posted by Chris Anderson on August 17, 2008 at 11:00pm
As Jordi continues testing and improving the ArduPilot boards, I've been integrating his cool binary-mode GPS parser and otherwise improving our pre-alpha code. So I think we're now a proper alpha, which is to say that although this hasn't been tested in the air yet, it does have most of the ingredients of a basic autopilot. The Arduino code is here. Improvements include: --Binary mode GPS parser with checksum error checking and high-speed (56k) throughput --Hardware-driven servo control (using the Arduino Servo Timer library), which means less processor overhead, tighter response and no jitters. --Now samples the rudder at autopilot initiation, so if the rudder is trimmed a bit one way or another that will be retained --Autopilot board LEDs now show GPS status We're going to do a few more alphas as we test the code on the latest versions of the hardware and against the simulator. Then, once we've tested it in the air, we'll move to beta. So consider this version just instructive and don't fly anything with it!
Read more…

Thermal accelerometers R back

Posted by Jack Crossfire on August 12, 2008 at 12:00pm
Spring mounted mass accelerometers truly do suck. Thermal accelerometers were once the standard & they're coming back now that everyone knows what high frequency vibration does to spring mounted masses.It's China to the rescue again, as the thermal accelerometers from Memsic are being used in thousands of persistence of vision wands at the olympics. The wands can display any communist party approved message when waved by manual labor. There are no pictures of the wands except for the thousands of blue dots in stadium wide shots. It must be an election year.
Read more…

EasyStar Project

Posted by Jacob Dickinson on August 11, 2008 at 4:00pm
I’m not going to be able to work on this project for awhile, so I thought I would post about it, partially so that I don’t forget the details myself : )For those of you who don’t know, I’m currently working on a UAV based on a Multiplex EasyStar and an Arduino Diecimila. My main goal currently is to simply get it to fly from waypoint to waypoint, but in the distant future my goal is to extent flight time as long as possible. I’ve been working on the project for the past couple months, and it’s my first foray into both RC and serious microprocessor programming. Currently, the navigation algorithm is extremely simple, and relies entirely on GPS – if it’s too low, go up, if it’s not on the right heading turn the rudder until it is. No accelerometers, no gyros. I know I’ll eventually have to add some sort of method for the plane to know its own orientation when I start doing serious flying, but for now I’m just trying to get all of the hardware collected and the basic software written.The navigation and GPS processing algorithms I wrote from scratch, using trigonometry and such. I believe that they’ll only work in the Western hemisphere, but I’m really not sure. I find the absolute heading to one point from another by calculating latitude and longitude components of the distance (over the curve of the Earth) from the one point to the other, then treating those distances as legs of a flat 2-d triangle. I can then find the angle from absolute North to the destination point. It’s probably almost too precise, but it’s the way that makes the most sense to me, and the processor seems to be able to handle it well. I’ve pretty happy with the results of car-based tests of the autopilot…the update rate of the entire system is about 3 Hz, which I think is fine for a slow, stable glider.The aircraft has never actually flown under autopilot control, but everything has been tested on the ground and seems to work. So, when I next get a chance to work on it, my first priority will be to clean up the rat’s nest of wires on the protoshield...then install something to keep it stable in the air and start testing. It’s probably going to be on hold until December, when the semester is finished, but I think I will have enough time then to get it working in the air.Pictures and code for those who are interested follow...cheers!

Haha it looks terrible...but prototypes never look clean. You can just barely see an Arduino Protoshield underneath...NavigationMain.zip
Read more…