Embedded Files
Design Performance and Cost Review with Alpha Prototype Demonstration
Design Performance and Cost Review with Alpha Prototype Demonstration
Subsystem 1 Alpha testing
Subsystem 1 Alpha testing
Subsystem 1 is a combination of the image detection, image recognition (neural network), and object tracking modules. This subsystem is one of two subsystems which together make up our total project.
Subsystem 1 is a combination of the image detection, image recognition (neural network), and object tracking modules. This subsystem is one of two subsystems which together make up our total project.
In order to test Subsystem 1, the following milestones were designated
In order to test Subsystem 1, the following milestones were designated
Rotate platform with SS1 Pi utilizing NEMA-17 motor
Design and machine stepper motor mounting bracket and belt tensioning device
Build up test code to rotate platform
Research motion tracking algorithms
Find an algorithm which runs efficiently on Raspberry Pis
Needs to be high FPS and run on top of image detection loop
Develop working image detection model
Take enough pictures of the Drones to develop proper model
Find preexisting model to retrain
Rotate platform with SS1 Pi utilizing NEMA-17 motor
Rotate platform with SS1 Pi utilizing NEMA-17 motor
Platform shown here mounts and tensions NEMA-17 stepper motor and rubber belt. This platform is designed as a structural addition to the existing platform and is sturdy enough to rotate the platform along with all the torque requirements of the fully built up system.
Platform shown here mounts and tensions NEMA-17 stepper motor and rubber belt. This platform is designed as a structural addition to the existing platform and is sturdy enough to rotate the platform along with all the torque requirements of the fully built up system.
Motor in motion responding to test code
Motor in motion responding to test code
The video shown here demonstrates control in both directions of both continuous and discrete steps. Utilizing both the AdaFruit Stepper Motor Hat along with the motorHat python library, the stepper motor is able to be controlled with a single command. This allows the motor control to be light weight and fit into the larger detection loop.
The video shown here demonstrates control in both directions of both continuous and discrete steps. Utilizing both the AdaFruit Stepper Motor Hat along with the motorHat python library, the stepper motor is able to be controlled with a single command. This allows the motor control to be light weight and fit into the larger detection loop.
IMG_2196.MOVIMG_1972.MOVMotion detection algorithm
Motion detection algorithm
The following video demonstrates an image tracking algorithms ability to place bounding boxes over moving objects in frame.
The following video demonstrates an image tracking algorithms ability to place bounding boxes over moving objects in frame.
Late stage testing of the Drone tracking software
Late stage testing of the Drone tracking software
In order to properly identify and track the drone, roughly 1000 training images were used to perfect the model. The following video demonstrates the systems ability to first identify the Drone within the frame, and track its movements across the screen.
In order to properly identify and track the drone, roughly 1000 training images were used to perfect the model. The following video demonstrates the systems ability to first identify the Drone within the frame, and track its movements across the screen.
drone_obj_det.mp4Subsystem 2 Alpha testing
Subsystem 2 Alpha testing
Subsystem 2 consists of the RF detection and deterrence systems. In the Alpha testing stage of this subsystem, the team is able to view the RF spectrum about the center frequency of the Drone's control signal in order to predict where in the window communication is occurring if a drone is present. In the future this system will be awoken by SS1 in order to conserve on power and system resources.
Subsystem 2 consists of the RF detection and deterrence systems. In the Alpha testing stage of this subsystem, the team is able to view the RF spectrum about the center frequency of the Drone's control signal in order to predict where in the window communication is occurring if a drone is present. In the future this system will be awoken by SS1 in order to conserve on power and system resources.
The following video shows the live Drone control spectrum in the top right highlighted in white. The code shown is a detection algorithm that tests the spectrum for Drone control signals and displays its best guess at where the drone communication window is at that time window. This positively demonstrates SS2's ability to detect a drone through RF interpolation.
The following video shows the live Drone control spectrum in the top right highlighted in white. The code shown is a detection algorithm that tests the spectrum for Drone control signals and displays its best guess at where the drone communication window is at that time window. This positively demonstrates SS2's ability to detect a drone through RF interpolation.
Locate_Drone_RF_Demo_1.mp4Page updated
Google Sites
Report abuse