Dynamic LED-based optical localization of a mobile robot / Jason N. Greenberg.

Autonomous mobile robots operating in areas with poor GPS and wireless coverage (e.g., underwater) must rely on alternative localization and communication techniques to navigate the field, share their data, and accomplish other missions. This dissertation is focused on the design of an LED-based optical localization system that achieves Simultaneous Localization and Communication (SLAC), where the bearing angles, needed for establishing optical line-of-sight (LOS) for LED-based communication between beacon (base) nodes and a mobile robot, are used to triangulate and thereby localize the robot. A two-dimensional (2D) setup is considered in this work.First, the measurement process and procedural steps necessary for implementing the localization scheme are developed. Critical to the success of this scheme is the maintenance of the LOS, which is difficult due to the robot's mobile nature. A Kalman filtering-based approach is proposed to predict the mobile robot's position, allowing the system to reduce the overhead of establishing and maintaining the LOS, therefore significantly improving the quality of the localization and communication. The effectiveness of this approach is evaluated with extensive simulation and experiments, including a comparison to an alternative approach not using Kalman filtering-based location prediction.The initial design of the localization system involves two base nodes, which could result in a singularity problem in position measurement when the mobile robot is close to forming a collinear relationship with the base nodes. To address this issue, a setup involving more than two base nodes is considered, where one could dynamically change the base node pair for localization. An important design consideration for this approach is how to best exploit the redundancy in base nodes to provide robust localization. A sensitivity metric is introduced to characterize the level of uncertainty in the position estimate relative to the bearing angle measurement error, to dynamically select a desired pair of beacon nodes. The proposed solution is evaluated with simulation and experimentation, in a setting of three beacons nodes and one mobile node, and its efficacy is demonstrated via comparison with multiple alternative approaches.The aforementioned work assumes that the bearing angles with respect to all base nodes are captured simultaneously (or when the robot is at a single location). Consequently, because scanning for the light intensity to determine the bearing angle takes time, a stop-and-go motion has to be used to ensure that the robot is at a single location during the angle measurement process, which significantly slows the robot's movement. To counter this issue, a scheme is proposed to dynamically localize a robot undergoing continuous movement, by exploiting the velocity prediction from Kalman filtering to properly correlate two consecutive measurements of bearing angles relative to the base nodes. Simulation and experiments show that, with this approach, the robot can be successfully localized when it is continuously moving.

Online resource; title from PDF title page (viewed on Feb. 14, 2022)