Accurate Path Tracking of Ground Mobile Robot for Simultaneous Localization and Mapping
HA XUAN VINH
Prof. Ha Cheol Keun
- 원문 URL
In trend of science development, mobile robots have been an essential element in the use of specific tasks, as exploration, building maps, surveillance, assistance, logistic services, demining, environmental monitoring, and others in an unknown environment. These tasks can be performed alone or with other devices in collaborating teams, with a various degree of autonomy, from the simple human supervisor remote-operated systems to a truly autonomous system capable to high level planning and decision making. Additionally, for intelligent mobile robots, they must determine their location and the environment map precisely. Therefore, the Simultaneous Localization and Mapping (SLAM) algorithm is considered as a basic capability of mobile robots necessary for autonomous navigation. The research effort in this field has thus increased over the past few years. In fact, the accuracy of the path tracking of ground mobile robot is very effective to perform the localization which is a part of SLAM algorithm. The sensor noises, which come from sensor such as Inertial Measurement Unit (IMU), incremental encoder, laser range finder (LRF) and etc… as well as the wheel slip phenomenon, lead to inaccurate estimation path of mobile robot. This dissertation focuses on the optimization algorithm of noise covariance parameters and wheel slip compensation in order to obtain accurate path tracking for SLAM of ground mobile robot in indoor environment. The first work in this dissertation is the combination of the Particle Swarm Optimization (PSO) and Mesh Adaptive Direct Search (MADS) algorithms, so-called PSO-MADS, in tuning Extended Kalman Filter (EKF) problem. In the robotic research, the Kalman filter (KF) is a well-known filtering algorithm to determine a robot's position based on sensor fusion technique. The performance of the filter depends largely on the knowledge accuracy of the process noise covariance matrix and the measurement noise covariance matrix. Several global optimization algorithms are applied to tune the EKF. However, these global optimization algorithms can easily get trapped in local optima when solving a complicated cost function, with a large number of local optima. The experimental results proved that the proposed PSO-MADS algorithm is an effective method to obtain more accurate estimated trajectories than the other algorithms, such as GA and PSO based EKF tuning. Next, based on the more accuracy of the robot position and heading by using the PSO-MADS algorithm, the wheel slip of mobile robot is investigated. Wheel slip limits the traction and braking ability of the robot when the robot is travelling on different ground surfaces with different and longer trajectory shapes. Therefore, a novel method for online wheel slip estimation based on a Discrete KF (DKF) to compensate for the velocity constraints is introduced. Additionally, the tuning fuzzy Vector Field Orientation (FVFO) feedback control, which is a control strategy, is developed by using the Fuzzy Logic Control (FLC) to tune parameters of the VFO method. The experimental results show that the novel wheel slip compensation method overcomes the limitations of the others in the trajectory tracking control problem. Finally, this work deals with a probabilistic SLAM formulation and it is focused on EKF-SLAM, which is the most frequently implemented SLAM framework. The use of the optimization and slip compensation methods above is very effective to obtain high accuracy of robot's path tracking and environment map for SLAM application in the mobile robot field.
- 이 논문과 함께 이용한 콘텐츠