Tight Formation Flight Research Articles

Structured H∞ synthesis: A systematic design approach for fully and underactuated systems

The task of designing a robust control system using structured H∞ algorithm involves several issues: choosing the system topology, the controller structure and tuning parameters to achieve robust stability and performance. Its application can lead to a trial-and-error approach at some stage. In order to minimize that effort, a methodology is proposed. A step-by-step guide is provided and includes modelling an uncertain system and designing roll-off, notch and setpoint filters aided by singular value analysis. The step of choosing a suitable topology for the control system and the controller structure is accomplished by RGA matrix analysis. Using a mixed-sensitivity approach, the design of weighting functions to achieve desired frequency-based performance is carried out considering plant restrictions. Application of the structured H∞ optimization algorithm is commented and results are validated by means of sensitivity function analysis and nonlinear simulation. A successful study case of control system for tight formation flight is presented which includes nonlinear disturbances, uncertainties, measurement noise, input saturation and restrictions due to RHP poles. Results are compared with those obtained with other robust control techniques.

Discrete Sliding Mode control of small UAS in tight formation flight under information constraints

This paper is concerned with a new control strategy based on discrete sliding mode control of small Unmanned Aerial Systems (UAS) in tight formation flight under information constraints. A discrete robust control strategy based on the sliding mode approach and a leader-follower scheme is proposed to achieve the desired flight performances while assuming realistic information constraints. A predictive reaching law is compared to a linear reaching law. This predictive discrete sliding mode control (PDSMC) strategy allows taking into account hard actuator constraints by solving an optimization problem at each time step.Also, this paper presents a meaningful study and comparison of the two discrete sliding mode control designs and time sampling continuous sliding mode control (TSCSMC). Here, the comparison focuses on the effect of discrete sampling on the control error analytically and in simulation. Effects on mesh stability are evaluated in simulation. Simulation results of a flight scenario with two different sampling frequencies demonstrate the efficiency of the proposed control strategy and show clearly the effect of the sampling time on the formation flight performance of the UAS obtained by the considered control strategies.

An approach to ground based space surveillance of geostationary on-orbit servicing operations

On Orbit Servicing (OOS) is a class of dual-use robotic space missions that could potentially extend the life of orbiting satellites by fuel replenishment, repair, inspection, orbital maintenance or satellite repurposing, and possibly reduce the rate of space debris generation. OOS performed in geostationary orbit poses a unique challenge for the optical space surveillance community. Both satellites would be performing proximity operations in tight formation flight with separations less than 500m making atmospheric seeing (turbulence) a challenge to resolving a geostationary satellite pair when viewed from the ground. The two objects would appear merged in an image as the resolving power of the telescope and detector, coupled with atmospheric seeing, limits the ability to resolve the two objects. This poses an issue for obtaining orbital data for conjunction flight safety or, in matters pertaining to space security, inferring the intent and trajectory of an unexpected object perched very close to one׳s satellite asset on orbit. In order to overcome this problem speckle interferometry using a cross spectrum approach is examined as a means to optically resolve the client and servicer׳s relative positions to enable a means to perform relative orbit determination of the two spacecraft. This paper explores cases where client and servicing satellites are in unforced relative motion flight and examines the observability of the objects. Tools are described that exploit cross-spectrum speckle interferometry to (1) determine the presence of a secondary in the vicinity of the client satellite and (2) estimate the servicing satellite׳s motion relative to the client. Experimental observations performed with the Mont Mégantic 1.6m telescope on co-located geostationary satellites (acting as OOS proxy objects) are described. Apparent angular separations between Anik G1 and Anik F1R from 5 to 1 arcsec were observed as the two satellites appeared to graze one another. Data reduction using differential angular measurements derived from speckle images collected by the 1.6m telescope produced relative orbit estimates with better than 90m accuracy in the cross-track and in-track directions but exhibited highly variable behavior in the radial component from 50 to 1800m. Simulations of synthetic tracking data indicated that the radial component requires approximately six hours of tracking data for an Extended Kalman Filter to converge on an relative orbit estimate with less than 100m overall uncertainty. The cross-spectrum approach takes advantage of the Fast Fourier Transform (FFT) permitting near real-time estimation of the relative orbit of the two satellites. This also enables the use of relatively larger detector arrays (>106 pixels) helping to ease acquisition process to acquire optical angular data.

Nonlinear Formation Guidance Law with Robust Disturbance Observer

Many formation guidance laws have been proposed for VAV formation flight. Since most autonomous formation flight methods require various active communication links between the vehicles to know motion information of other vehicles, damage to the receiver or the transmitter and communication delay are critical problem to achieve a given formation flight mission. Therefore, in this point of view, the method that does not need an inter-vehicle communication is preferred in the autonomous formation flight. In this paper, we first summarize the formation guidance law without an inter-vehicle communication using feedback linearization and sliding mode control proposed in previous study. We also propose the modified formation guidance law with robust disturbance observer, which can provide significantly better performance than previously mentioned guidance law in case that other vehicles maneuver with large accelerations. The robust disturbance observer can estimate uncertainties generated by acceleration of leader vehicle. By eliminating the uncertainties using the estimated uncertainties, VAVs are able to achieve the tight formation flight. The performance of the proposed approach is validated by numerical simulations.

Tight Formation Flight Control

Thetightformatione ightcontrolproblemisaddressed.Theformationconsistsofa lead andwingaircraft,where the wing e ies in tight formation with the lead, such that the lead’ s trailing vortices aerodynamically couple the lead and the wing, and a reduction in the formation’ s induced drag is achieved. A controller (i.e., a formation-hold autopilot for the wing aircraft ) is designed such that the formation’ s geometry is maintained in the face of lead aircraft maneuvers. In the formation e ight control system, the wing and lead aircraft dynamics are coupled due to kinematic effects, and, in the case of tight formations, additional aerodynamic coupling effects are introduced. These additional aerodynamic coupling effects are properly modeled. The most signie cant aerodynamic coupling effect introduced by tight formation e ight entails the coupling of the lateral/directional channel into the altitudehold autopilot channel. It is shown that formation-hold autopilots designed ignoring the aerodynamic coupling effect yield satisfactory performance in tight formation e ight. Nomenclature b = wingspan of wing CLL = lift coefe cient of the lead aircraft S = surface area of wing VSW = sidewash W = wash vector WUW = upwash