Decentralized control and estimation are both active research areas in the field of systems and control. A new approach to these topics utilizes graph theory to characterize inter-agent communication as a graph that, in this thesis, can have time-varying topology. This approach has been named "network-decentralized" and the use of network-decentralized control and estimation can enable multi-agent systems to achieve tasks in low information environments. In this thesis a novel network-decentralized control algorithm is proposed to enable collision avoidance and formation producing behavior in a multi-agent system. Additionally, a network-decentralized estimation algorithm is also proposed that is combined with the network-decentralized control approach yielding a first of its kind network-decentralized network-estimated control algorithm. A multitude of simulation environments are developed to test the algorithm in a 2-D holonomic multi-agent environment and a parameter tuning method is presented. Strong performance is shown in 2-D with collision avoidance guarantees, at the cost of difficult to tune parameters. The use of holonomic agents in 2-D is consistent with most research in decentralized control and estimation. Since this type of agent is common, this thesis extends its work to the 3-D non-holonomic agent case in an attempt to help characterize how accurate, scalable, and useful existing decentralized research is for real world applications. To develop the network-decentralized algorithm to this point, it is first expanded to the holonomic 3-D case where it is further tested in newly developed simulation environments, where again, collision avoidance guarantees are provided. A novel free-space metric is introduced which allows for tests to be compared across dimensions (eg: 2-D and 3-D) and also yields addition insight as to when parameter-performance breakdown will occur. For the extension of the algorithm to a non-holonomic vehicle in 3-D a suitable quadcopter platform is designed. In addition, the Delft Center for Systems and Control (DCSC) Distributed Robotics Lab is re-designed and completely virtualized to allow for students to easily access lab resources, such as motion capture (MoCap) information, quadcopter test-code environments, and documentation all in one location. A first-principles based system identification process is presented and used to identify the designed quadcopter platform so that a suitable controller can be found for in-lab use. Two quadcopter simulation environments are developed to allow for the design of hover-envelope controllers based on identified parameters. Since the DCSC department was unable to acquire multiple physical vehicles for swarm testing the system identification process and quadcopter simulation environments are validated against the test results from a single physical quadcopter in the newly designed lab environment. A more comprehensive multi-agent implementation of the collision avoidance portion of the algorithm is tested in a real-time vehicle simulation environment. It is found that while the single physical agent can closely track the behavior of holonomic agents and quadcopter simulation results, the real-time multi-agent case exhibits a degradation in performance that further declines in obstacle dense environments. This explains a loss of performance that can lead to collisions in the real-time setting where in the holonomic case the exact same test yields successful results.