Loading...
Enhancement of Mobile Robot Navigation and Localization
Renawi, Abdulrahman M.
Renawi, Abdulrahman M.
Date
2017-11
Author
Advisor
Type
Thesis
Degree
Citations
Altmetric:
Description
A Master of Science thesis in Mechatronics Engineering by Abdulrahman M. Renawi entitled, "Enhancement of Mobile Robot Navigation and Localization," submitted in November 2017. Thesis advisor is Dr. Mohammad A. Jaradat and thesis co-advisor is Dr. Mamoun Abdel-Hafez. Soft and hard copy available.
Abstract
Nearly all mobile robots need the knowledge of the robot’s location with respect to the initial position or with respect to the control station position to be able to navigate. Many research projects are conducted to solve the localization and navigation problems, but they are all either specific for indoor or outdoor due to the limitations of the available localization sensors or with some precautions and delays for indoor-outdoor localization with switching in-between. This thesis presents a single low-cost approach for the enhancement of localization and navigation of wheeled mobile robots for indoor-outdoor environments. Using a ZED Camera, an Inertial Measurement Unit, a Global Positioning Sensor and wheels Encoders, filtered with a Kalman Filter with no switching, a trajectory controller is designed and implemented to guide the robot through experiments based on a system model, and it is proved to be stable and efficient. Simulations are carried out on a Gazebo simulator using a Robot Operating System, and field experiments were done on a Kobuki robot platform to validate the proposed controller. Sensors’ measurements are fused using Extended Kalman Filter and Unscented Kalman Filter due to the system’s non-linearity. Filters’ results are simulated on Matlab using field data which is proved to be stable. Then, the results are implemented to fuse sensors’ readings onboard and estimate the robot’s location in the local frame. Both filters show good accuracy with 0.13 meters in indoor experiments, 0.2 meters in outdoor experiments, and 0.6 meters in indoor-outdoor experiments. The Unscented Kalman Filter shows lower absolute true ground error values than the Extended Kalman Filter by 0.01 meters. The proposed approach is efficient for indoor, outdoor and indoor-outdoor scenarios. Indoor and outdoor results were outstanding. Indoor-outdoor experiments show promising results. In addition, Unscented Kalman Filter outperforms Extended Kalman Filter in true errors.