New IARC Presentation Videos!

Look for them under the videos page.

Notes from IARC 2009 – continued…

Individual Paper Presentations:

Low Cost Guidance Navigation Presentation (Gerogia Tech)

• Video Link
• Claim to be first to use Matlab and RC transmitter together for rapid controls development

Embry Riddle

• Video Link
• Small camera
• xbee, most processing offboard

AUVSI Rankings for Best Robotics Schools

• Contact Director of Knowledge, invite to Waterloo

MIT

• (See presentation videos on the videos page)
• One propeller/motor upside down, to clear way for camera. Rest are right way up, supposedly to keep centre of pressure high relative to centre of mass
• 1kHz operation, noisy IMU
• Trouble with visual odometry because of texture free walls, rely on lasers

Oregon State University

• Video Link
• I2C speed controller with 500 kHz refresh rate

PIMA

• Blimp

Georgia Tech 2nd Presentation (IA entry)

• Video Link
• Paper Link: An ILS Inspired Approach and Departure System Utilizing Monocular Vision by Christman and Johnson
• Similat to M. Tribou’s work?
• Uses LEDs, semi directional, low power

Other Notes

MIT, Georgia Tech and Embry Riddle are the only universities are the only teams flying this year

Lots of wind inside, because of narrow hallways

Obstacles like pedestal fans, chairs and hanging items from the cieling

Walls are paper

Georgia Tech Observations

• Team comprised of PhD’s, grad students and one professor
• 3 presentations, 2 of which were technical papers
• Simple, self stabilizing co-axial helicopter
• In aggressive maneuvers the two blades touch each other and break, happened very often during testing runs
• The arena at UPR has lots of sky lights, not exactly an IR friendly environment and the metal frame prevents any useful GPS use
• The common Sharp IR range sensors gave them problems
• The sonar altitude sensor, saw a chair obstacle as an increase in ground hieght, almost went through the roof, had to be intervened by a judge

MIT Observations

• Collaborated with a German startup (Ascending Technologies)
• Almost made it on the third try but the LIDAR saw the flapping wall and the localization failed. IARC judges reinforced the ‘wall’ with a metal frame for their next run

VT and OSU Observations

• Programmed and tested their quadrotors single axis at a time
• VT had a basic PID control structure with 8 analog sonars, 3DMG, will add LIDAR next year, used low frequency phoenix ESC’s
• OSU had a PD control loop with a high frequency I2C ESC

 

 

 

Notes from IARC 2009

While we maintained a presence at the IARC this year, we did not compete. The competition took place in Puerto Rico in July.

We will gradually upload notes from various presentations, conversations and observations. Feel free to discuss in the comments!

Notes from Michelson’s Opening Presentation:

Possible MAV missions:

1. Home into radars, swarms, disrupt air traffic

2. Home into radios and do damage, disrupt communications

3. Prime mission: underground facilities, something that flies in, sits in a powered down state and activates as needed, stealthy

Current Problems:

1. Scaling, not only Reynolds #, but also wavelength of communications frequencies

2. Indoor: communication problems and GPS problems. Navigation needs to be autonomous. Can’t have a high gain GPS antenna on a small scale. Similarly a scaled down com antenna won’t talk through walls.

Examples: Black widow, Mirador

 3. Weight: Bulk of energy going to propulsion

4. Latency: eliminating teleoperation would eliminate a number of sources of latency

5. Energy

6. Aperture: Wide bandwidth video feed conflicts with high frequency to accommodate size of vehicle

Example: bump on global hawk is a satellite, aperture size of communications

7. Self Stability: Can assume cultural environment like vertical walks and horizontal roof, need a reference in absence of GPS, maintain starting memory

8. Avoidance: Reactive, Preplanned, random decisions to avoid getting stuck

Indoor Navigation without GPS

1. Ultrasonic system, own GPS system inside, before Novatel and other companies people didn’t know how easy GPS was and we became reliant on GPS thereafter

2. How do birds do it? Vision, Gravity, Range estimation (stereo vision), Acoustic cues, Airflow direction, Friendly or foe / potentially useful obstacle detection

 Low Power Communications and Sensing in Spectrally Cluttered Environment

1. Noise Sources

2. Multipath

3. Near field spectral clutter, e.g. if trying to implement a radar, how do you interpret your reflections

4. Low power transmission @ high frequency

Indoor Fixed Wing

1. Not good

 e.g. Aerovironment UAV – could get sucked into a vent, 90 s flight time, landing problems, can’t go so fast

2. If you can get 5 minutes inside a building you can cover a lot of rooms

3. If can land then can power down and maybe plant a sensor, mission can last weeks. E.g. land and wait for someone to turn on lights.

Tail Sitter MAVs for Indoor Flight

1. Efficiency not so great, some possibilities

Lighter Than Air

1. Toy blimps, but 15 cm is impractical: only 1.84 grams of lift

2. In a breeze it would act like a sail

3. Drag

Quadrotors

1. Many teams

2. Around for a while, 1923

Co-axial Helicopter

1. Better for indoor

2. Fly slowly

3. Highly controllable

Flapping Wing

1. Hard for controls

2. Everything in nature flaps wings except maple seeds?

e.g. Entomopter, Aerovironment Humming bird, CIA dragonfly

Conclusions:

1. Close proximity to obstacles

2. Spectrally cluttered multipath

3. No teleops, No GPS

4. flapping wing etc good

 

Stay tuned for more notes and videos!