Abstract
Recent technical advancements in both fields of unmanned aerial vehicles (UAV) control and artificial intelligence (AI) have made a certain realm of applications possible. However, one of the main problems in integration of these two areas is the bottle-neck of computing AI applications on UAV's resource limited platform. One of the main solution for this problem is that AI and control software from one side and computing hardware mounted on UAV from the other side be adopted together based on the main constraints of the resource limited computing platform on UAV. Basically, the target constraints of such adaptation are performance, energy efficiency, and accuracy. In this paper, we propose a strategy to integrate and adopt the commonly used object detection and tracking algorithm and UAV control software to be executed on a heterogeneous resource limited computing units on a UAV. For object detection, a convolutional neural network (CNN) algorithm is used. For object tracking, a novel algorithm is proposed that can execute along with object tracking via sequential stream data. For UAV control, a Gain-Scheduled PID controller is designed that steers the UAV by continuously manipulation of the actuators based on the stream data from the tracking unit and dynamics of the UAV. All the algorithms are adopted to be executed on a heterogeneous platform including NVIDIA Jetson TX2 embedded computer and an ARM Cortex M4. The observation from real-time operation of the platform shows that using the proposed platform reduces the power consumption by 53.69% in contrast with other existing methods while having marginal penalty for object detection and tracking parts.
Highlights
Object recognition and tracking is one of the most challenging tasks in autonomous aerial vehicles since the detection and tracking the objects should be accurate and agile in run-time and with rational energy consumption
To verify the feasibility of the proposed algorithm, several experiments were performed under variable speed of the detected target to obtain the optimal values for the parameters gains that is used for object-tracking algorithm
The quadcopter is flying at an altitude of 2 meters, once the RC car is detected, the real-time object detection is triggered and a boundary box is drawn around the moving RC car based on our proposed object detector
Summary
Object recognition and tracking is one of the most challenging tasks in autonomous aerial vehicles since the detection and tracking the objects should be accurate and agile in run-time and with rational energy consumption. A heterogeneous platform for stream data processing in real-time object detection and tracking is proposed. Two heterogeneous onboard processing units are used, i.e., big and little for object detection and following, and the tasks are managed on them based on the requirements and constraints of the system. While the little processing unit is responsible for tracking process Using this heterogeneous platform and appropriate scheduling of tasks, we adjust the performance and accuracy of the tracking part according to the requirements of the feed-back based system. The output of the object tracking algorithm is sent to the flight controller (little processing unit) to start the tracking process by sending the required pulse width modulation (PWM) values to the motors.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
