I'm still a little fuzzy on the advantage of the "maple-seed" style monocopters, but BotJunkie has an update on Lockheed Martin's Samurai UAV:
"We heard some rumors back in 2009 that Lockheed Martin’s SAMARAI UAV project had been killed off. In fact, Lockheed Martin themselves apparently confirmed that they had stopped developing the UAV since AeroVironment won DARPA’s nano air vehicle development contract to put some polish on their robot hummingbird. So, I’m not entirely sure what the background to this video is, but it shows a much more recent (and smaller, with a wingspan of only 12 inches) version of the SAMARAI.
The SAMARAI is certainly simple (at least, compared to AeroVironment’s UAV), and as the video shows, you can just chuck it into the air, and it can land on the ground and then take off again without needing much in the way of space or infrastructure. On the other hand, I’m not sure exactly how you’d go about mounting something like a camera on a spinning airframe (maybe sync the shutter speed with the rotation speed?), and in order to operate effectively indoors, the SAMARAI would benefit from some level of resilience to impacts. At this point, it seems as though a collision with a wall or doorframe would probably knock the SAMARAI a tad askew, causing it to spin out of control and decapitate everyone in the room.
LabVIEW based ArduPilot Ground Control Station ( GCS) is ready for public beta test. GCS wiki: See wiki for installation and configuration instructions (read through it, or at least watch video )
Requirements:
- Windows OS - LabVIEW run time engine 9.0 - Google Earth ( if you already have it you don't have to re-install it. Google Earth version prior to 5.1 ( newest) should be fine)
This tiny drone is able to do completely automatic flights, it can perform altitude hold (based on sonar sensor) and obstacle avoiding based on IR distance sensors (you could see the "black stange eyes" on the photo). It´s your personal droid...
Look at the video (the "tennis game" part it´s funny. Thanks to Ramon for the idea!!)
There are some new features in this thrid part... This is the list:
For outdoor configuration:
- GPS library support (actually UBLOX or NMEA)
- Position hold based on GPS
For indoor configuration:
- 4x IR distance sensors to detect obstacles (1.5m range)
- Obstacle avoiding (using distance sensors)
Common:
- Altitude hold based on Sonar (LV-EZ0)
- Automatic flight pattern (experimental).
--- Automatic takeoff
--- Position hold [outdoor] or obstacle avoiding [indoor] during a predefined time
--- Automatic descend
--- Automatic landing
- Added XBee for telemetry (and debug)
And some improvements in the code:
- New "radio test mode" to test radio equipment
- Revised control routines
Development
For the GPS position hold I had to implement the navigation algorithms for the quadcopter because it´s really different that the one used for planes...For this navigation it´s necesary to have the magnetometer to cancel the yaw drift in hover conditions. One thing I have observed is that you can only fly this tiny drone on very calm days because it´s too light for the wind... so it´s better suitted as an indoor drone. Then I started to think how to make a cheap way to navigate on indoor enviroments... I have one sharp IR disntace sensor so I start making some tests mounting the sensor in a servo to make a 180º scan. The idea was to mount 2 (or 4) of this sensors in the moving head. On the tests I found that in this little machine the moving head caused some inestability, so I decided to mount 4 sensors in a fixed way. OK, this the cheap DIY version of an EXPENSIVE laser range finder, but it works... there are many thing to improve and test, but it´s a promising start...
Details
Sonar module is an LV-EZ0. Because we don´t have any analog input available I use the PWM interface in a Port Change pin (PCINT20) to use an interrupt to read the sensor. (It´s recommended some solder skills to make this modification).
For the IR range finder (Sharp GP2Y0A02) I needed to use a separate Arduino Pro mini (again we don´t have any analog input free). This module connects to the ArduIMU via Serial port so we need to choose between GPS of range finder (outdoor-indoor decision).
On this III part, the hardware (ArduIMU) really show it´s limits... it´s not a problem of CPU power, it´s a problem of the limited I/O as I said before, so it´s time to move to the big brother, the newArduPilot MegaHardware... this new platform will be fantastic for this projects...
Behind the scenes
During the test of position hold I have some crashes (nothing important, only some broken propellers...) and there was a moment in that the quad performs not so good, so I start searching the reason. Again I suspect that it could be a vibrations problem so I decided to make a modified code to test the vibration on each motor.
As you can see I have problems on left motor, so I change this prop, also add a new layer of doubled sided foam tape to the ArduIMU and problem gone.
Respect to the IR distance sensor, the first version was a moving head with a servo but this had some problems with vibrations that affect stability and also has a poor scanning rate, here is a photo of this prototype. Finally I decided to use 4 fixed sensors.
Codes
Some parts of this codes are still experimental but you can get it here:
For the very few of you who were interested, here are some more pictures from our project, which when compared to the hundreds of amazing things you guys do on this site, seems pretty low:
In case anyone's wondering, the material we use is twinwall, kind of like the plastic version of cardboard (two plastic sheets connected by flutes running between them). The thing is, I know they're used alot for SPADs but not so much for slightly more complex things - is there anything about twinwall we've massively overlooked?
I know this has taken ages (we only get an hour a week to work on this :( ) but we've nearly finished the wing and the fuselage, but there's still so much to do - at least it'll keep me occupied for the next few months at least.
A company that won a NASA prize for beaming power to a climber in a space-elevator competition is targeting unmanned aircraft as an initial application for its laser-based technology.
LaserMotive says power beaming could extend the endurance of electrically powered UAVs. The company plans to fly a small internally funded demonstrator by year’s end.
I changed the motor driver to a l298 to provide more amps to my motors than the l293d did. (l298 picture below)
I am back from testing my AGV some more. Here is some video of it going around a retention pond in a parking lot. The first gps point was a little off and it came close to the grass but it wasnt too bad. Almost ready for Sparkfun's contest lol.
typical GPS precision plus turbulence gives around 10-15m precision.
After spending one minute on manual adjustments:
This is what you have after importing automatically, without any manual adjustments, IMU corrections included but the major factor is that the camera is already almost vertical and the platform is stable:
Since the protos are taken with roll-stabilised head using Pteryx UAV,
they are ortorectilinear as the shooting angle was at worst a few deg relative to vertical.
FLEXIPILOT log decoder now take camera lens angle and mounting offset angles as a parameters, generating Google Earth file with matching photos within seconds! This is particularly useful for isolated shots, since overlapping images when mapped in quantity give an impression of 'randomized crowd'.
Another approach: Microsoft ICE and 10min processing after drag & drop:
Unfortunately the image is a little skewed.
The major benefit of stabilised head is that all photos are of acceptable geometry and useful for stitching. However, because the precise moment of taking the photo cannot be known, an important angle and positional error will always be present in the logs. Another factor to include is changing lighting and altitude, inevitable in real-life operation and during turbulent weather (this time 42km/h flight speed and 20km/h wind speed plus strong thermals affecting pitch).
Automated mapping (AerialRobotics LogDecode) applied to a linear map:
"At ICRA last week, Stanford presented some footage of the latest trick they’ve been able to teach their robotic Volkswagen Passat. Driving backwards at 25 mph, Junior autonomously hits the brakes and starts to skid, at the same time rotating 180 degrees to (most of the time) end up neatly in a parking space. The video shows a couple different techniques: “closed-loop control” means that Junior is performing on the fly adjustments to its trajectory based on sensor data, while “open-loop control” means that Junior is simply executing a set of commands that result in a powerslide motion. Ideally, the entire maneuver would be closed loop, but the dynamics of a powerslide are too complex to model accurately. So instead, Junior uses closed-loop control during the lead-in backwards acceleration phase, and then transitions into open-loop control as the slide starts.
Now, part of the reason that Volkswagen/Audi are helping to fund all of this stuff is that eventually, they hope to use the technology in production automobiles to help keep us all safe. And, if you think about it, there are those occasional times where you see a parking spot open up behind you on the opposite side of the street with someone else about to snag it. By implementing this parking technology in consumer cars, you’ll be able to park that much faster, thus decreasing the amount of time that you’re on the road and therefore lowering your accident risk. Autonomous powerslide parking: it keeps you safe!"
It is a great fact that UAVs tend to miniaturization. Despite the high technology advances on microprocessors and embedded systems it is clear that only simple control laws can be implemented into Miniature Autonomous Vehicles. Note that i am not talking for rf manned control but for fully autonomous solutions. So what about Wireless Sensor Network control for Drones? It is an idea implemented widely in ground robots. The main approach is to implement on-board only the main control functions (for example stabilization with simple PID loops for safety mainly when the wirelless network is not availible for some reasons) and let advanced control laws send commands from the ground station. Obvioysly the drone will also send its state vector through the network. What's your opinion?
The folks over at Grass Roots Mapping are looking for people with low altitude aerial photography skills to help map the BP oil spill in the Gulf of Mexico.
Some interesting thinking in this week's Robots Podcast, on the past 50 years and next 50 years of robotics. One of those interviewed is Jean-Christophe Zufferey, who focuses on UAVs. Along with discussing the past two decades' advances in UAVs very well, he speculates on what's next, including the idea of "personal UAVs", like flying assistant droids, which could give you an aerial view of your surroundings or do communications relay.
He also discusses the possibility of "3D elevators" or transportation UAVs, that would allow commuter aircraft without the danger and complexity of having to drive. You'd just get in, give it a destination, and get out when you got there.
Adafruit has a released a very handy new datalogging shield ($19). You can attach any sensor and it saves the data to a SD card. Use it with the standard Arduino dev board.
The included Real Time Clock timestamps all your data with the current time, so that you know precisely what happened when!
Please note that this item does not come with an Arduino (you'll need one to use with the shield), or an SD card. It does come with the RTC battery, however. The kit is un-assembled, You'll need some basic soldering skills to put it together, but even if you don't have much experience you can get it done in under 1 hour.
SD card interface works with FAT16 or FAT32 formatted cards. 3.3v level shifter circuitry prevents damage to your SD card
Real time clock (RTC) keeps the time going even when the Arduino is unplugged. The battery backup lasts for years
Included libraries and example code for both SD and RTC mean you can get going quickly
Prototyping area for soldering connectors, circuitry or sensors.
Onboard 3.3v regulator is both a reliable reference voltage and also reliably runs SD cards that require a lot of power to run"
Inspired by the Sparkfun Autonomous Vehicle Competition, where massive points were given for autonomous takeoff and landing, we're going to do something like that here for round 7. So here's the DIY Drones T3 Total Autonomy Round--"Look Ma, No Hands!"
Rules:
Contestant who takes off autonomously, flies autonomously and lands autonomously closest to their take-off spot---TWICE--wins. Distance from final rest and launch should be averaged over both runs. It doesn't matter what the aircraft does between launch and landing. One loop is fine.
Contestants must submit KML track AND video. Just set up your video camera on a tripod viewing your launch/landing spot. We want to see: A) your transmitter on the ground. B) the plane landing as close to possible to the launch spot. C) You looking very relaxed, doing nothing. Extra points for you in a lounge chair with a drink.
Entries must be submitted in the comments here by 12:00 midnight PST on Sunday, July 4th. Winners will, as always, get points for the cumulative T3 leader board and some cool prize TBA. Remember to tell us what autopilot and aircraft you used.
FAQ:
Q: What do you mean by "take off"? Is hand-launching allowed?
A: Yes. As long as your RC transmitter is on the ground, you can hand launch
Q: Why twice?
A: So you don't crash your plane into the launch spot and call that an "autonomous landing".
Q: What does T3 stand for?
A: Originally "Trust Time Trial" but since this round doesn't involve time, it's just for continuity. Trust is still operative, however.
Q: Is this an awesome time to try out the new autonomous takeoff and landing feature in ArduPilot 2.6?
A: Why yes, it is!
[Picture of Jordi Munoz not doing an autonomous landing at the Sparkfun AVC.]
This week's guest is Missy Cummings, the director of MIT's Humans and Automation Lab and one of the first female navy fighter pilots.
I've already edited and prepared the podcast but my internet isn't fast enough to upload it. Tomorrow I'll be able to upload it (probably well before the evening).
To better show the potential of (DIY) flexcircuits, here is a board that includes the wiring bit and board to board connection. There are 14 connections to the mother board (8 in use).
Very nice plans to make a smallish Predator out of bluecore foam. Too small for a decent UAV (and as we've said here many times, Predators are not the best flyers, being both short-coupled and prone to twitchy pitch, and vulnerable to tip-stalling on landing). But if you want to scale this up, these plans are excellent and would be a good place to start.
Today was the first day to get a glimpse of the sun behind the clouds. So i used the chance for a video session and in depth testing of the new Catamarane hull (with the fins).
I tried various speeds, various payloads and various tracks to see if the new platform is stable.
It was!
1. A new speed-record with 10km/h (running with 8A and a 500g Li Battery-Pack)
2. The big battery-pack with about 1.5kg weight was also OK and had only a slight impact on the speed (about 1km/h slower). With that pack, the endurance of the ship will be around three hours with one charge.
(running with about 5A and a speed of about 6-7 km/h).
3. The straight-line stability on long runs was also OK, but still has some optimization potential.
After that runs, i had enough courage to mount my beloved Sony DigiCam (a DCR P100) with some gaffer-tape on the ship. I made two runs, the first with the cam attached to the front and the second with the cam attached to the aft. The aft-footage gives a nice look, how the ardupilot is controlling the yaw-servo.
All went well.
Despite the fact that the ship had a near-accident with one of the diving teachers of our club, who managed to surface at exactly the moment, i launched the ship.
Maybe we can make a bet, how much lawsuits for "malicious injury" i will have collected until the end of the year.
Under this title there is an article in the Weekend section of The Times (UK) newspaper today with a quarter page picture of Chris Anderson and Daniel.
