Data Acquisition

Data acquisition will be used to control the drone during flight. The data will also be stored to be used for Professor Raman’s research to be processed at a later date. 

Locust Sensor

Image from Professor Raman’s lab

Professor Raman has developed a sensor that can be implanted into a locust in order to measure its muscle movements and brain signals. This sensor will be used to collect data during flight in order to explore the locust’s brain signals when exposed to positive and negative stimulus via odor plumes. We were unable to implement this sensor during our project due to COVID-19.

Parrot Mambo

The Parrot Mambo comes with an assortment of sensors that will be used to control the drone and monitor its states. It is equipped with an ultrasonic sensor, two cameras, accelerometer, and gyrometer. The Parrot Mambo is also capable of measuring battery voltage which will be useful for system identification and control. All of these sensors are collected in real time and stored on the drone. This data can be downloaded post flight using the associated MATLAB and Simulink packages for post processing and graphing

OptiTrack

The OptiTrack System is located in Green 0161 on Washington University’s campus. It is comprised of 16 cameras and is able to track reflective surfaces in a 3D space. This data can be transmitted in real time to the drone for use in guidance and navigation as well as stored for later use. By adding reflective balls to the drone, the OptiTrack System can be used to track the drone. Streaming location data from the OptiTrack system to the drone requires additional hardware and software (shown above).The OptiTrack program is hosted on a designated computer in Green 0161. The data is then streamed to another computer via LAN or WiFi. A MATLAB script was written to record this data on the new computer. By using TCPIP (transmission control protocol/ internet protocol) the position and angle data can then be transferred to the drone for feedback. An example of the position as well as Euler angle data can be seen below.

This system was employed at the beginning of the project. It was proven that data could be collected in real time and and transmitted to the drone.

Kalman Filter

Due to the number of imperfect sensors used in this project, state estimation was required. A steady state Kalman Filter (Wiener Filter) was implemented. MATLABS LQE (linear quadratic estimation) function was used in order to calculate the ‘L’ matrix shown before.

This was very effective at eliminating noise and small biases from our sensors. Measured noise was added into the system during simulation in order to simulate the drone as accurately as possible. A few example observed and actual states can be seen below.