Stereo Vision Implementation For Aerial Surveillance using Autonomous Aerial Vehicle
The project presents a solution for autonomous aerial surveillance using quadcopter. It includes the development of complete working strategy for aerial surveillance, capable of continuous perimetral surveying in a low flying or industrial environment. In such an environment, the first requirement f
2025-06-28 16:36:09 - Adil Khan
Stereo Vision Implementation For Aerial Surveillance using Autonomous Aerial Vehicle
Project Area of Specialization Electrical/Electronic EngineeringProject SummaryThe project presents a solution for autonomous aerial surveillance using quadcopter. It includes the development of complete working strategy for aerial surveillance, capable of continuous perimetral surveying in a low flying or industrial environment. In such an environment, the first requirement for autonomous maneuvering is a robust obstacle avoidance strategy. Here, stereo vision is presented as such a strategy. Other features that are required for autonomous surveillance are object detection and tracking.
The purpose of this project is to save manpower and provide autonomous low-cost security solution for perimetral surveillance. In this project, depth estimation is performed through stereo vision and test other object detection and tracking algorithms.
This project mainly includes three tasks which will be running in parallel. Main task of this project is autonomous flight of aerial vehicle following the specific way points. Instead of receiving commands from transmitter and then moving in that direction, motion of quadcopter will be decided depending on the GPS coordinates provided by user. For that, current position of quadcopter will be determined and then comparing it with the next position, the direction of motion is decided and required signals will be generated.
To move a quadcopter autonomously, it is necessary to develop an obstacle detection and avoidance system to avoid any kind of collision in case any obstacle comes in the way of flight. So, stereo vision based obstacle avoidance is used in this project as it is one of the most effective methods. Stereo cameras are being used to capture the video on run time and then processed by raspberry Pi to calculate the disparity of images being captured by both of the cameras. Using the output disparity, distance of the objects in each frames will be calculated and then in case of any object closer than 1m, signals will be generated to move on one side to avoid the obstacle.
Live video will be being transferred to computer for surveillance. For transfer of video and other information, wifi network will be used. To track an object, object will be selected on computer by creating region of interest around it. It will provide the pixel position of object in that frame. This information of position of object and the frame will be sent to quadcopter and that will use this information to detect the position of object in new frame being captured as there will be a delay due to transferring of data. Once the new position of object is detected, tracking program will start and will control the quadcopter to keep the position of the target object in the center of frame. If the object is lost or an abort instruction is received, tracking will be stopped and quadcopter will start following the way points.
A graphical user interface will be developed to make it easy for an operator to select region of interest using cursor or send an abort command.
Benefits of the ProjectWith the advancement in technology, world is moving toward replacing human beings with the autonomous systems as these are less expensive and better to you in many cases. This autonomous drone will provide surveillance of a place while flying over the specified trajectory. Surveillance can be done continuously and there will not be need of any person to control the drone. There will only be need of one person to watch over the videos being streamed from the drone and send required commands. Hence, it will need less manpower. Surveillance of an area can be done from a distant or remote place and this place may not be accessible by humans due to any condition such as radioactivity etc. Surveillance of such areas can be done easily using an autonomous drone and it will be cost efficient as well. Apart from the surveillance, it can be used to search for a lost person or an object by just providing the way points of an area you want to search for and then just look at the monitor to find it.
If someone or something breach the security, it is needed to keep track where ever it goes. This project provides an easy solution for the tracking of an object if operator finds it suspicious or object of interest. Hence, the object will be tracked and operator will be getting the position and status of the object so the necessary action can be taken. This can also be used by police to track any person or vehicle.
Technical Details of Final DeliverableThe final product of this project will include a quadcopter and a control unit which will be a computer in this case. Quadcopter will be able to fly on a specific height and with specific speed. It will be following the way points provided either when it was programmed or received from the control unit. Stereo cameras installed on front of the quadcopter will be capturing the live video. This video will be processed and it will be if there is detect any obstacle closer than 1 m. If obstacle is detected, quadcopter will be moved either on left or right to avoid obstacle. This video will be being transferred to the control unit using the wifi network. When any information is received regarding tracking of object, quadcopter will stop following the way points and will start tracking the required object. Once quadcopter receives any command to abort the tracking or the object being tracked is lost, quadcopter will again start following the way points and start moving to the next coordinate. The signals from remote control will always override the all other operations. This is to make sure that if any problem occurs in autonomous control system, quadcopter can be controlled manually.
On the side of control unit, live video of quadcopter will be being received. A graphic user interface will be provided to the operator to control the quadcopter. Using the GUI, new way points will be sent to the quadcopter so a new trajectory can be followed. During the course of flight, if any object is being noticed which is needed to be detected, operator will have option to draw a rectangle to select region of interest surrounding object to be detected. This information will be sent to quadcopter with the command to follow the target object. if it is needed to abort the tracking, there will be an option in GUI to send a command to quadcopter to abort the tracking task and start moving along the way points.
Final Deliverable of the Project Hardware SystemCore Industry SecurityOther Industries IT , Transportation Core Technology RoboticsOther Technologies Internet of Things (IoT)Sustainable Development Goals Industry, Innovation and InfrastructureRequired Resources| Item Name | Type | No. of Units | Per Unit Cost (in Rs) | Total (in Rs) |
|---|---|---|---|---|
| Total in (Rs) | 55300 | |||
| Quadcopter Frame QAV250 Glass Fibre | Equipment | 1 | 2500 | 2500 |
| Flight Controller OmniBus F4 v3 | Equipment | 1 | 4000 | 4000 |
| Motor RS2205 with 30A Electronic Speed Controller | Equipment | 4 | 4250 | 17000 |
| Propeller GF5045 BN | Equipment | 4 | 200 | 800 |
| Transmitter/Receiver FlySky | Equipment | 1 | 10000 | 10000 |
| Stereo Camera ELP Dual Lens 2560x1920 | Equipment | 1 | 13000 | 13000 |
| Lipo Battery Vant 3s 4200 mah | Equipment | 1 | 8000 | 8000 |