Target Tracking Implementation on Embedded Platform

Project Title

Target Tracking Implementation on Embedded Platform

Project Area of Specialization Artificial IntelligenceProject Summary

Demand for AI based embedded system is growing in embedded domain. With artificial Intelligence based advanced trackers, the accuracy of the target detection and tracking has drastically improved in the continuous environment. We aim to implement target detection and tracking for Anti-Aircraft Machine Gun (AAMG) for Heavy Industries Taxila (HIT). This will be an important contribution to the department of defense since this technology has not being deployed yet. Our objective is to achieve efficient real time target detection and tracking system capable of monitoring motion of target under dynamic environmental conditions.

This project is to introduce automated tracking mechanism in the AAMG and to shift the trend from manual tracking to automated tracking. The automated tracking system will offer ease to target the enemy also provide safety to the gunman. Our system is easy to implement on any military tank or armored vehicles with proposed customizations and upgradation as per requirements. Artificial Intelligence based techniques are to be adopted and incorporated with the feedback control system of motors which would track the target. Our target is to introduce this technology to the military of Pakistan viz; Armed forces, Rangers, Police official. We also look forward to export the technology to friendly countries resulting in increase of foreign exchange also enhancing country’s prestige.

Project Objectives

Main Objective

Main objective of our project is to achieve efficient real time target detection and tracking system capable of monitoring motion of object under dynamic environmental conditions.

 Sub-Objectives

1. To introduce a real time automated object tracking and targeting system in the 12.7mm Anti-Aircraft Machine Gun (AAMG) and to shift the trend from manual tracking to automatic tracking.
2. Port a real time tracking to embedded based platform.
3. Saving the precious life of the gun man.
4. Contribution to the defense system of Pakistan.

Project Implementation Method

In this project first we implement the target tracking on the PC and then ported it on the embedded platform. The following are the steps of project implementation.

1. Set up the environment and software installation on the PC.
2. Run various implementations of Siamese tracker on PC to compare various results.
3. Set up embedded platform and integrated it with LCD, IP camera and motion control card.
4. Run the Siamese tracker on the embedded board to track an object.
5. Get the coordinates of the object location and transfer them to the motion control card to move the turret through motors to track the object.

The following diagram shows the tracking process.

           

Benefits of the Project

The project provides the following benefits.

1. Contribution to the defense system of Pakistan.
2. Less burden and safety of the tank operator.
3. Any type of customization is possible as it is being built in house.

Technical Details of Final Deliverable

An IP camera is connected to the turret, this camera is connected with the computer through a LAN cable. To track an object motors are used to move the turret in horizontal and transverse position. Signals are sent to motors though motor drive. This motor drive is connected with the computer through serial communication module. Motor drive accept the analog voltage (variable voltage 0 to 5V) or PWM to control the speed and direction of motors. To test the platform, it is tested with 9 variable speeds in both directions to track object (results are not accurate) with a 6 bit and (digital to analog converter) DAC.

To track an object in the video BOOSTING, MIL, KCF, TLD, MEDIANFLOW, GOTURN, MOSSE, CSRT and Kalman filter-based trackers are tested. They track the selected object and send the signal to motor drive so that they rotate the motors towards the selected target.

All of these algorithms are decade old and work ok in controlled environment. Therefore, Neural Networks based advanced trackers are available and are popular for improved performance in various environments and backgrounds. Although Neural Networks based techniques are more computationally expensive than traditional algorithms but the results obtained from it have best-in-class performance.

           

The commander in the tank selects the targeted object and Artificial Intelligence based algorithm tracks the object. In which network is trained on various datasets and any object can be tracked through this algorithm. AI based tracker determine the current position of the object in the live camera feed, when current position of the target is determined errors are calculated and speed of the both motors are predicted through PID (Proportional integral derivative) controller so that the gun moves toward the target. If the motors are reached at desired position they stopped. Again, if the target moves motors move in that direction.

Final Deliverable of the Project HW/SW integrated systemCore Industry SecurityOther IndustriesCore Technology Artificial Intelligence(AI)Other Technologies RoboticsSustainable Development Goals Industry, Innovation and InfrastructureRequired Resources

Item Name	Type	No. of Units	Per Unit Cost (in Rs)	Total (in Rs)
			Total in (Rs)	80000
NVIDIA Jetson Board	Equipment	1	70000	70000
Miscellaneous	Miscellaneous	5	2000	10000