Group Members:
Weiyun Jiang, Alexis Yang, Ning Du and Jiajun Wan
I. ANNOTATED BLOCK DIAGRAM
II. EXTERNAL BEHAVIORAL SPECIFICATION
Elephant-Human Conflict (EHC) is aimed at solving the problem of elephants invading human living areas for crop-raiding in African and Asian countries. Our system consists of drones for scaring elephants away from human living areas and GPS tags that are placed on elephants to allow tracking. The behavioral specifications are listed in the order of our demonstration.
An XBee transmitter is installed at a ground base. This transmitter just acts as a beacon that sends out a signal constantly. The function of this transmitter is to activate any GPS module that is within the communication range, about 1 km.
The user needs to install the GPS tags on the adult male elephants around a given village. The GPS tag consists of a GPS receiver and an XBee RF module to send out the GPS coordinates received when the tag is within 1km of the ground base. The GPS receiver is in sleep mode most of the time to save power.
The Xbee on the GPS tag constantly tries to receive the signal from the ground base. When the tag comes within 1km of the ground base according to signal strength from the Xbee in the ground base, the microcontroller in the tag turns on the GPS receiver in the tag and starts to transmit the GPS signal to the XBee on the drone.
The drone receives the coordinate broadcasted by the tag and flies to the vicinity of the GPS coordinate.
When the drone is close enough to the elephants based on the GPS signal. The Drone’s object detection and tracking system turns on and attempts to actively follow the elephants and scare it off.
The camera on the drone constantly detects the elephant using Yolo V3 once it gets close to the elephant and the sound of the drone propellers will scare off the elephants.
Once the elephants are scared away from the ground base, the XBee signal sent from the ground base becomes more and more attenuated. When the signal from the ground station received by the tag dip below a certain value, the tag automatically turns off the GPS and stops sending coordinates to the drone.
When the drone stops receiving the GPS signal as the tag turns off the GPS transmission, the drone returns to the ground base.
The process repeats itself after the tagged elephant is again within 1km range of the ground station.
III. MEMBER RESPONSIBILITIES
Since our 3-way communication network and embedded system on the drone will take a substantial amount of time, the physical drone construction will add huge complexity to our project. Most importantly, a working, stable drone is crucial to later computer vision tasks, so we would like to reuse the hexacopter from the past project Eternal if possible.
Drone:
After the physical drone is constructed, the majority of the tasks fall on the computer vision object detection algorithm and the flight control algorithm to track the elephants. To be more specific, the Yolo V3 algorithm implementation and the tracking algorithm by flight control using the Onboard SDK are the two essential parts. First of all, once the camera detects elephants using Yolo V3, an algorithm to keep the elephants frames inside the view of the camera as the elephants move needs to be implemented. In addition, this part also includes integration of the Pi Camera, the Xbee module, and interfacing TX2 with the N3 flight controller. Overall, the algorithm development and testing is a large task, so at least two group members will be responsible for the drone system.
Tag:
A tag consists of a GPS module and a Xbee module. The GPS module serves to obtain the longitude, latitude, altitude, and velocity information from the satellites. On the other hand, the Xbee module functions to transmit any valid information to the drones. The Xbee module on the tag not only needs to transmit GPS information, but also should look for the signal from the ground station. If the Xbee module of the tag receives any ground station signal, the elephant with the tag will be regarded as dangerous and the GPS module on the tag will be turned on. In the end, the arduino prototype is expected to be integrated on a single PCB to minimize the size. Two group members will work on the tag.
Weiyun Jiang: Arduino PCB Design, Object Detection and Tracking
Alexis Yang: Object Detection and Tracking & TX2 Peripherals Configuration
Ning Du: Xbee Communication and GPS
Jiajun Wan: DJI SDK Development and Flight Control