Abstract
Nonholonomic vehicles are typical dynamic systems in engineering application. Formation control of multiple vehicles aims at making a team of vehicles move with a featured group motion and maintain a desired geometric pattern. This topic has attracted tremendous attention since the last decade, due to the rapid development of unmanned vehicle systems. In practice, vehicle formations are able to complete complicated tasks, lower miscellaneous costs and improve working efficiency. Though much progress has been made on formation control of multiple nonholonomic vehicles in the past few years, yet velocity constraints, absence of global information, limited local measurements, and complicated collective motions still give rise to numerous challenges. This thesis focuses on the leader-follower parallel formation and circular formation control problems of networked nonholonomic vehicles of unicycle type. The network among vehicles is set up by onboard sensors and/or communication devices.The leader-follower parallel formation control problem is one of the fundamental formation control problems. The follower vehicles are required to follow the leader’s motion while maintaining a desired geometric structure. The first part of the thesis addresses the leader-follower parallel formation control problems of vehicles subject to the velocity constraints described by saturated angular velocity and bounded linear velocity lying between two positive constants. The network topology is modeled by a directed graph. The leader-follower formations defined in the inertial frame and in local coordinate frames of vehicles are considered respectively. The problem formulation includes the consensus of multi-vehicle systems as a special case. The main results of this part is summarized as follows:
1. The case where the desired vehicle formation is defined in the inertial frame, is studied. The proposed dynamic control law for each follower only uses its local information and the information of its neighbors in the network. It is shown that the leader-follower parallel formation can be achieved without using absolute position measurements while the velocity constraints are satisfied.
2. The case where the desired vehicle formation is defined in the local coordinate frames of vehicles, is considered. Each vehicle is assumed to have its local Frenet Serret frame and thus a common reference direction is not required. It is shown that in two cases, the follower vehicles can converge to the leader’s motion with a desired formation, while the velocity constraints are satisfied.
In the second part of the thesis, circular formation of nonholonomic vehicles with respect to a static point is investigated. The objective is to design a distributed control law such that a group of vehicles can travel along a common circle and maintain a spaced formation. In practice, circular formation can be applied to enclosing, capturing, securing or monitoring a target. Each vehicle is assumed to have its local coordinate frame and be subject to the velocity constraints. The main results of this part is summarized as follows:
1. The case where the network topology is modeled by a cycle, is studied. The proposed dynamic control law guarantees that all vehicles can globally converge to a common circle with the given static center and radius, and maintain an evenly spaced formation along the circle with a given steady-state velocity. The proposed control law does not rely on any global information or common settings.
2. The case where relative distance measurements and the communication among vehicles are not available, is studied. A static controller is developed such that vehicles can globally converge to a circular formation around a given center with any desired spacing, provided that the radius of the common circle and the steady-state velocity are known and a certain subgraph of the sensor graph contains no directed cycles.
3. The case where the network topology is modeled by a general directed graph, is studied. The proposed dynamic control law guarantees that all vehicles can globally converge to a given circular formation with any desired spacing. The proposed control law only requires intermittent communication and can be applied to the circular formation control problem of dynamic unicycles.
In the final part of the thesis, circular formation control of networked nonholonomic vehicles with respect to a moving target, is considered. The moving-target circular formation control problem is more complicated than the one with a static target, in the sense that vehicles additionally need to track the target as it moves. Each vehicle has its local coordinate frame and the topology of the sensor graph may dynamically switch based on the relative positions between vehicles and the target. A static controller using only local measurements of each vehicle is proposed. It is shown that the communication among vehicles is not required and a reconfigurable evenly spaced circular formation around a moving-target can be achieved.
| Date of Award | 2 Dec 2016 |
|---|---|
| Original language | English |
| Awarding Institution |
|
| Supervisor | Lu LIU (Supervisor) |