Collision avoidance of UAVs using stereo camera

• Main Author: Tang, Tee Yang
• Other Authors: Wang Dan Wei
• Format: Final Year Project
• Language: English
• Published: 2018
• Subjects: DRNTU::Engineering::Electrical and electronic engineering::Control and instrumentation::Robotics
• Online Access: http://hdl.handle.net/10356/75170
• Institution: Nanyang Technological University

Unmanned aerial vehicles (UAVs) market is projected to grow rapidly due to its versatility, costeffectiveness, and technological advancement. Most UAVs depend on global positioning system (GPS) to accurately locate its position. However, such UAVs are unable to operate in the urban and indoor environment where the GPS signal is degraded or denied. Alternative localization methods involve fusion of different sensor data reading, namely laser rangefinder, VICON motion sensor, and visual sensor with an inertial measurement unit (IMU). Recent development in visual sensor makes it a viable choice for various UAV applications because of its low cost, low power consumption, and light weight. Extensive research on sensor fusion between visual sensor and IMU, also known as visual inertial odometry (VIO), have been carried out in recent years to improve the accuracy of UAV localization. The Extended Kalman Filter (EKF) to estimates the state of UAV by computing multiple measurements over time from various sensors while considering the individual sensor statistical noise. On top of the localization issue, most indoor UAVs navigate at a specified altitude and map out 2D obstacle map, thus limiting its maneuverability. An efficient 3D mapping framework, OctoMap, can be used to overcome the issue of mapping and navigating in the 3D environment. Additionally, obstacle avoidance algorithm is required for UAVs to safely navigate in an unstructured environment. Thus, other obstacle avoidance algorithms such as RRT and PRM have been evaluated and ranked according to their computational efficiency. The objective of this project was to develop a low-cost and lightweight UAV that can navigate in semi-autonomously in the GPS-degraded/denied 3D environment. The proposed objectives for the UAV, in terms of cost and weight, are less than S$1,000 and 2 Kg respectively. In an effort to lower the cost of UAV, open-source software and hardware such as Linux operating system, robot operating system (ROS), Odroid XU4, and SP Racing F3 Deluxe flight controller have been chosen. The system, once developed, is fundamental for search and rescue and urban warfare application. However, the lack of computational power and the inherent properties of visual odometry limits the functionality of the developed indoor UAV.