When a mission profile of an unmanned micro air vehicle is known a priori, one of the strategies adopted in autonomous control is to first generate a compatible trajectory in off-line and then implement a controller to track the trajectory. However, in decision rich intelligent aerial robots and in 6DOF operations of unmanned aerial vehicles, the mission profiles are usually known during an operation. Hence the trajectories are instantaneously inferred. Here an adaptive controller with capabilities to track these trajectories is required. In this paper, it is shown that an extended Kalman filter is an excellent tool to design such a controller which accepts the trajectory generation module as an input and then reconstructs its trajectories. To illustrate this feature, a 3DOF micro air vehicle in pitch plane is considered and an adaptive controller (referred as autonomous control agent) using an extended Kalman filter is applied to infer typical adaptive mission profiles.
INTRODUCTIONAutonomous control of aerial robots in 6DOF is quite challenging. Unlike unmanned aerial vehicles for surveillance and reconnaissance operations, the adaptive mission profiles of aerial robots are usually inferred during its operation. Consequently, the trajectories are instantaneously inferred. When an adaptive control law is required to track these trajectories, design and implementation procedures using a conventional controller is handicapped. For instance, consider a known mission profile. Compatible trajectory in offline is generated and a human-made conventional controller is implemented online so that the trajectory is tracked to achieve the known mission profile. This approach has been demonstrated through flight testing [1,2]. To adopt these techniques for instantaneously generated trajectories belonging to an adaptive mission profile, an autonomous control agent (similar to an adaptive controller) with capabilities to reconstruct the instantaneous trajectories are required. In this paper, a feasibility study is performed to show that extended Kalman filter (EKF) is an excellent tool to design the autonomous control agents (ACAs). In this case, the EKF algorithm [3], in state and parameter estimation framework is used. That is, it takes the trajectory generation module as an input (measurements) and delivers an ACA to reconstruct its trajectories sequentially such that an adaptive mission profile is realized. Suppose a linear state feedback structure is used, the gains of the ACA become time-varying and the closed loop system becomes non-autonomous. In this case, stability is verified by checking the eigenvalues of the piecewise linear system. ACA presented in this paper fills in the adaptive controller layer conceived early in the development of a multi-layered autonomous control [4]. These layers also consist of trajectory generation, trajectory reconstruction, trajectory planning modules [5,6]. Adaptive controller emphasized in this investigation reconstructs the trajectory generated through waypoints [5]. Since ...