This week was heavily focused on sensor research. We needed to know everything about sensors that might be useful to us in the future. I centered my research on accelerometers and ultrasonic sensors. I took notes on my research and shared them with the rest of my group. Those notes can be viewed here:
https://docs.google.com/a/erhsnyc.net/document/d/1IDusj62JcVtWjOUq0Q2_YyC9YIXPkqSnfm2RgsoXeBU/edit
Next week will be focused on both the Linux SDK and the cameras in the drone. Specifically, I will see, along with other members of my group, how the cameras work with the SDK.
- Guilherme Dias
Monday, March 9, 2015
Week of 3/2/15
AR Drone
Notes:
- Quadrotor UAV with each opposite pair or rotors turning clockwise, the other turning counterclockwise
- Indoor and outdoor configurations
- Powered by brushless engines with three phases, microcontroller
- Automatic detection of blade stopping, stops the all rotors
- Sensors consist of many sensors below the hull
- 6 DOF, MEMS-based inertial units used for pitch, roll, yaw
- Ultrasonic sensor for altitude control
- Camera aiming towards the ground/front facing 90 degrees
- 3 DOF with 3 axis magnetometer and pressure sensor
- Tilt sensors used for hovering and stabilization
Accelerometer:
- electromechanical device used to measure acceleration forces
- movement (vibration) of the sensor creates voltage, which changes the output
- how they work:
- piezoelectric effect: microscopic crystals that get stressed by accelerative forces → creates voltage
- can have two microstructures next to each other → they have capacitance between them → when one moves, capacitance changes → can convert capacitance to voltage using circuits
- can analyze tilt and movement of the drone
- can sense if the drone is falling/crashing
- drone can react if it falls (security measure)
- analog vs digital (based on hardware)
- analog: outputs voltage continuously in proportion to acceleration
- digital: will produce a square wave; acceleration determines amount of time for which the voltage is high
- for 3d movement, we need either a 3 axis accelerometer or two 2 axis accelerometers mounted at right angles
- for a fast moving machine, we need an accelerometer with a larger bandwith (amount of times per second we can get a reliable reading)
Ultrasound Sensors:
- measure properties of sound waves with frequency above human audible range
- based on: time of flight → senses distance; Doppler effect → senses velocity; attenuation of sound waves (combined effect of scattering and absorption (conversion of sound energy to other forms of energy)) → sense directionality
- can detect clear or shiny objects that other sensors do not pick up
- measurements are extremely sensitive to temperature and angle of target
- if transmitting and receiving transducer are used next to each other, the distance calculated will be twice the actual distance
- transducer takes some time to switch between modes (sending and receiving), which causes a problem for short distance measurements
Gyroscopes:
- https://www.youtube.com/watch?v=cquvA_IpEsA - Basic info
- A gyroscope is a device with angular momentum used to capture orientation
- Gyroscopes maintain its orientation in space
- Used by aircrafts to control orientation
- Possible Gyroscope sensors can be used, rather than the physical
- Gyroscope will keep its orientation, resisting any outside force to change
- Consists of a central rotor within a gimbal
- Gyroscopic Precession https://www.youtube.com/watch?v=GeyDf4ooPdo
Magnetometers:
- Two purposes: “to measure the magnetization of a magnetic material like a ferromagnet, or to measure the strength and, in some cases, the direction of the magnetic field at a point in space”
- “Since magnetic flux density in air is directly proportional to magnetic field strength, a magnetometer is capable of detecting fluctuations in the Earth's field.”
- Magnetometers detect distortions in the magnetic flux which i
- 3 axis magnetometer 6° precision
- “With a 3D magnetometer, the AR.Drone 2.0 knows its precise orientation with respect to the smartphone, which becomes the reference point. The pilot no longer needs to care about the orientation of the AR.Drone 2.0's front camera, which will accurately track the smartphone's motion and tilt.
- ‘Relative Flight’ mode = conventional flight mode. This disables Parrot AR.Drone 2.0's magnetometer. The pilot manages the quadricopter's orientation with no assistance.”
- tells the drone precisely how it is oriented respective to the control device
- measures magnetic flux density at point where sensor is located. The max. distance the magnetometer can detect the sensor is directly proportional to the cube root of the magnetometer's sensitivity
*****
NOTES:
TUMS FLYING THROUGH A STRAIGHT LINE AND AVOIDING OBSTACLES-
****
Thursday, January 22, 2015
Week of 1/19/15
Progress: There hasn't been much progress lately (see Problems section). We are still setting up the spiri simulator and the container is ready for printing, although that has not been done yet. However, we found an open source program called OpenCV (Open Source Computer Vision)* that may be able to help with image recognition using the camera that is built into the Spiri, so that the drone can navigate throughout the school. The software seems to have an extensive library available in many languages, as well as a forum in which anyone can post questions and ask for help. This seems to be promising, but we need to try it out before we can determine if it is good enough for what we want.
Problems: Our group is not very well organized and the project is moving along very slowly. This is a major issue that we need to fix very soon.
Plan: In this coming week, we will print the container, fine-tune it, and hopefully finalize it so that the project can move along. Also, we will experiment with OpenCV to see whether is satisfies our needs or not. Furthermore, we hope to have the spiri simulator running sometime this week so we can start learning how it works and have a working demonstration of both the simulator and Gazebo by the end of the week.
Meanwhile, we will attempt to acquire the drone, and since the dev team is not responding, we may need to just buy it as soon as possible. Moreover, Gui will continue to learn the basics of SLAM on the side in case we need to turn to that option, but that would be an extreme case, because it is both difficult to learn and implement.
*OpenCV website: http://opencv.org/
Problems: Our group is not very well organized and the project is moving along very slowly. This is a major issue that we need to fix very soon.
Plan: In this coming week, we will print the container, fine-tune it, and hopefully finalize it so that the project can move along. Also, we will experiment with OpenCV to see whether is satisfies our needs or not. Furthermore, we hope to have the spiri simulator running sometime this week so we can start learning how it works and have a working demonstration of both the simulator and Gazebo by the end of the week.
Meanwhile, we will attempt to acquire the drone, and since the dev team is not responding, we may need to just buy it as soon as possible. Moreover, Gui will continue to learn the basics of SLAM on the side in case we need to turn to that option, but that would be an extreme case, because it is both difficult to learn and implement.
*OpenCV website: http://opencv.org/
Subscribe to:
Posts (Atom)