… in simulation for various failure cases. This paper is concerned with the development of an adaptive controller for an autonomous helicopter … for modeling … on small … EricN. Johnson?and Suresh K. Kannany School of Aerospace Engineering, 270 Ferst Drive, Georgia Institute of Technology, Atlanta, GA 30332-0150 cr reference model dynamics dderivative des desired fforce hhedge i;o inner loop, outer loop lat;lon lateral, longitudinal mmoment pproportional rrobustifyingterm, reference model II. Introduction

Unmanned helicopters are versatile machines that can perform aggressive maneuvers. This is evident from the wide range of acrobatic maneuvers executed by expert pilots. He- licoptershavea distinct advantage over flxed-wing aircraft especially in an urban environment, where hover capability is helpful. There is increased interest in the deployment of autonomous helicopters for military applications, especially in urban environments. These applications include reconnaissance, tracking of individuals or other objects of interest ina city, and search and rescue missions in urban areas. Autonomous helicopters must have the capability of planning routes and executing them. To be truly useful, these routes would in- cludehigh-speed dashes, tight turns around buildings, avoiding dynamic obstacles and other required aggressive maneuvers. In planning1 these routes, however, the tracking capability of the?ightcontrol system is a limiting factor because most current control systems still do not leverage the full?ightenvelopeof small helicopters, at least, unlesssigniflcant system identiflcation and validation has been conducted. Although stabilization and autonomous?ight2 has been achieved, the performance has generally been modest compared to a human pilot. This may be attributed to many factors, such as parametric uncertainty (changing mass, and aerodynamic characteristics) , un- modeled dynamics, actuator magnitude and rate saturation and assumptions made during control design itself. Parametric uncertainty limits the operational envelope of the vehicle to where control designs are valid, whereasunmodeled dynamics and saturation can severely limit the achievable bandwidth of the system. Theefiectof uncertainty and un-modeled dynamics have been successfully handled usinga combination of system identiflcation3{5 and robust control techniques.6 Excellent?ightand simulation results have been reported including acrobatic maneuvers7 and modestly aggressive maneuvers.6,8 3of 49 Adaptive Trajectory Control for Autonomous Helicopters, JOHNSON and KANNAN
A key aspect in the efiective use of unmanned aerial vehicles(UAVs) for military and civil applications is their ability to accommodate changing dynamics and payloadconflgurations automatically without having to rely on substantial system identiflcationefiorts. Neural- Network (NN) based direct adaptive control has recently emerged as an enabling technology for practical?ightcontrol systems that allow online adaptation to uncertainty. This tech- nologyhasbeen successfully applied to the recent U.S. Air Force Reconflgurable Control for Tailless Fighter Aircraft (RESTORE) culminating in a successful?ight demonstration9,10 of the adaptive controller on the X-36. A combined inner-outer loop architecture was also applied for guidance and control of the X-33 (Ref. 11) and evaluated successfully in simulation for various failure cases. This paper is concerned with the development of an adaptive controller for an autonomous helicopter usinganeural network as the adaptive element. For autonomous helicopters, a primary objective is the accurate tracking of position commands. Much adaptive control work on helicopters has concentrated on improving the tracking performance of attitude commands. 12{14 Usually a simple outer loop employing basic relationships between attitude and linear acceleration is then used to control the translational dynamics. For many applications this maybe su-cient. However, when operating in an urban environment or?ying information with other UAVs, the position tracking ability of the controller dictates the minimum proximity between the UAV and objects in its environment. In contrast to previ- ousattitude control-only work, we introduce a coupled inner-outer loop adaptive design that can handle uncertainty in all six degrees of freedom. In synthesizinga controller (Fig. 1), the conventional conceptual separation between the inner loop and outer loop is made. The inner loop controls the moments acting on the aircraft by changing the lateral stick, - lat , longitudinal stick, - lon and pedal, - ped , inputs. The outer loop controls the forces acting on the aircraft by varying the magnitude of the rotor thrust using the collective - coll input. The thrust vector isefiectively oriented in the desired direction by commanding changes to the attitude of the helicopter using the inner loop.

Download Adaptive Trajectory Control for Autonomous Helicopters.Pdf