Hojjat A. Izadi

Fault Tolerant Model Predictive Control of Quad-Rotor Helicopters with Actuator Fault Estimation

Abstract Model predictive control (MPC) at each time step minimizes a cost function subject to dynamical constraints to obtain a stabilizing control signal. Further, MPC is one of the few methodologies that can be used to design feedback control for nonlinear dynamical systems taking into consideration of actuator saturations. It can thus serve as a suitable fault tolerant control approach for quad-rotor helicopter governed by nonlinear dynamics. However, MPC needs a relatively accurate model of the post-failure system to calculate a stabilizing control signal. The problem becomes more critical where the system dynamics is described by a nonlinear model, because there exist few effective nonlinear parameter estimators with reasonable online computation time. To address this issue, for online actuator fault estimation, this paper investigates Moving Horizon Estimation (MHE) and Unscented Kalman Filter (UKF) as two methods for nonlinear parameter estimation. A framework is then formulated for integrating MHE/UKF based fault estimator with MPC to form an active fault tolerant control system for systems with nonlinear constrained dynamics. Performance and computation requirement of both algorithms are also investigated.

Hierarchical Decentralized Receding Horizon Control of Multiple Vehicles with Communication Failures

This work presents a new approach for designing decentralized receding horizon controllers (DRHC) for cooperative multiple vehicle systems with inter-vehicle communication delays arising from communication failures. Using DRHC each vehicle plans its own state trajectory over a finite prediction time horizon. The neighboring vehicles then exchange their predicted trajectories at each sample time to maintain cooperation objectives. Such communication failures lead to large, inter-vehicle communication delays of exchanged information. Large inter-vehicle communication delays can potentially lead to degraded cooperation performance and unsafe vehicle motion. To maintain desired cooperation performance during faulty conditions, the proposed fault-tolerant DRHC architecture estimates the tail part of the neighboring vehicle trajectory that is unavailable due to communication delays. Furthermore, to address the safety of the team against possible collisions during faulty situations, a fault-tolerant DRHC is developed, which provides safety using a safe protection zone called a tube around the trajectory of faulty neighboring vehicles. The radius of the tube increases with communication delay and maneuverability. A communication failure diagnosis algorithm is also developed. The required communication capability for the fault-diagnosis algorithm and fault-tolerant DRHC suggests a hierarchical fault-tolerant DRHC architecture. Simulations of formation flight of miniature hovercrafts are used to illustrate the effectiveness of the proposed fault-tolerant DRHC architecture.

Decentralized Receding Horizon Control for Cooperative Multiple Vehicles Subject to Communication Delay

N this paper, a new approach is proposed for the decentralized receding horizon control (DRHC) of multiple cooperative vehicles with the possibility of communication failures leading to large intervehicle communication delay. Such large communication delays can lead to poor performance and even instability. The neighboring vehicles exchange their predicted trajectories at each sampletimetomaintainthecooperationobjectives.Itisassumedthat the communication failure is partial in nature, which in turn leads to large communication delays of the exchanged trajectories. The proposedfault-tolerantDRHCisbasedontwoextensionsofexisting work for the case of large communication delays. The first contributionisthedevelopment ofanewDRHC approachthat estimates thetrajectoryoftheneighboringvehiclesforthetailoftheprediction horizon, which would otherwise not be available due to the communication delay. In this approach, the tail of the cost function is estimated by adding extra decision variables in the cost function. A relatively small amount of existing work has investigated the implementation issues associated with exchange of trajectory information, but so far no work has proposed a tail estimation process to compensate for large delays. For instance, in [1–3], no prediction or estimation for the trajectory of neighboring vehicles is performed, and it is assumed that the neighboring vehicles remain at the last delayed states broadcasted by them. Such assumptions may yield poor performance for large communication delays because the constant state vector is not a good estimation of a trajectory of states in general. Similar issues are also investigated in [4,5]. The second contribution of this paper is an extension of the tubebased model predictive control (MPC) approach [6,7] for the case of thelargecommunicationdelaysinordertoguaranteethesafetyofthe fleet against possible collisions during formation control problems. The concept of the tube MPC [or tube receding horizon control (RHC)] in existing work [6,7] is normally used to calculate a robust bound on the states due to system uncertainty, whereas in this paper, the approach is used to calculate bounds that arise from large communication delays of the exchanged neighbor trajectories. The proposed algorithms in this paper are presented in the context of fault-tolerant control, as the communication delay/break may occur due to any failure and malfunction in the communication devices. Some examples of communication failures for the team of cooperative vehicles can be found in [8–10]. In [8], the wireless communicationpacketloss/delayisconsidered;oncethepacketloss/ delay occurs, the previous available trajectory of the faulty unmanned aerial vehicle (UAV) is extrapolated to predict the future reference trajectory. Also, in [9], the communication failure in formation flight of multiple UAVs leads to a break in the communicated messages that forces the fleet to redefine the communication graph. This paper is organized as follows. Section II deals with a general formulation of the decentralized receding horizon controller, and the corresponding algorithm for a fault-free (delay-free) condition. In Section III, a faulty condition is first defined, and a reconfigurable fault-tolerant controller is developed. A safety guarantee method forthefaultyconditionisalsodevelopedbasedontheconceptoftube RHC. In Section IV, the proposed algorithms are tested through simulation of a leaderless formation controller for a fleet of unmanned vehicles.

Decentralized Model Predictive Control for Cooperative Multiple Vehicles Subject to Communication Loss

The decentralized model predictive control (DMPC) of multiple cooperative vehicles with the possibility of communication loss/delay is investigated. The neighboring vehicles exchange their predicted trajectories at every sample time to maintain the cooperation objectives. In the event of a communication loss (packet dropout), the most recent available information, which is potentially delayed, is used. Then the communication loss problem changes to a cooperative problem when random large communication delays are present. Such large communication delays can lead to poor cooperation performance and unsafe behaviors such as collisions. A new DMPC approach is developed to improve the cooperation performance and achieve safety in the presence of the large communication delays. The proposed DMPC architecture estimates the tail of neighbors trajectory which is not available due to the large communication delays for improving the performance. The concept of the tube MPC is also employed to provide the safety of the fleet against collisions, in the presence of large intervehicle communication delays. In this approach, a tube shaped trajectory set is assumed around the trajectory of the neighboring vehicles whose trajectory is delayed/lost. The radius of tube is a function of the communication delay and vehicles maneuverability (in the absence of model uncertainty). The simulation of formation problem of multiple vehicles is employed to illustrate the effectiveness of the proposed approach.

A Variable Communication Approach for Decentralized Receding Horizon Control of Multi-Vehicle Systems

The stability and performance of decentralized receding horizon control (RHC) is often improved by modifying the cost function and constraints of the problem. However, variable communication presents an additional approach for further improvement. In this paper, new methods are developed for variable communication of decentralized RHC. Through modification of stability conditions recently developed for decentralized RHC, an investigation of the effect of faster communication rates, expanded communication radius, and multi-hopping communication on decentralized RHC is performed. The new approach is applied to a group of five mobile robots with different communication topologies.

Decentralized receding horizon control of multiple vehicles subject to communication failure

In this paper, the decentralized receding horizon control (DRHC) of multiple cooperative vehicles with the possibility of communication failure is investigated. The neighboring vehicles exchange their computed trajectories at each sample time to maintain cooperation objectives. It is assumed that the communication failure is partial in nature, which in turn leads to large communication delays. A new reconfigurable DRHC approach is developed that guarantees the safety of the entire fleet in the presence of inter-vehicle communication failures. The concept of tube RHC is introduced to guarantee the safety of the fleet against collisions during faulty conditions. In this approach, a tube shaped trajectory set is used instead of a trajectory for the neighboring vehicles experiencing the communication failure.

A data-driven fault tolerant model predictive control with fault identification

Most of the existing active control methodologies need a post-fault/failure model of the faulty process for online retuning the controller parameters, or reconfiguration. However, post-fault model identification process takes the precious post-fault time which delays the recovery procedure. A new data-driven fault tolerant model predictive control (MPC) is developed which does not need the post-fault model. In fact, the model identification and control (re)calculation are combined together and are performed simultaneously to efficiently use the critical post-fault/failure time. The proposed fault tolerant architecture is capable of the online fault identification and adapting effectively to the post-fault/failure model. Several simulations of hover control of an unmanned quad-rotor helicopter are performed to illustrate the usefulness of the proposed approach.

Decentralized receding horizon control using communication bandwidth allocation

The decentralized receding horizon control (DRHC) of a team of cooperative vehicles with limited communication bandwidth is considered. It is well known that the more communication the better stability and performance properties of the cooperative vehicles; however, in reality the available communication is often limited. This motivates our research to develop a new algorithm for efficient usage of available communication capacity so that the teaming behavior is optimized. The proposed algorithm uses a bandwidth allocation method; the key idea is to reduce the overall mismatch between predicted and actual plans of each neighbor by efficient communication bandwidth allocation.

Rule-Based Cooperative Collision Avoidance Using Decentralized Model Predictive Control

A rule-based decentralized model predictive control (DMPC) approach is employed to address the collision avoidance problem of multiple moving vehicles. Every vehicle uses model predictive control (MPC) to plan its trajectory towards its assigned target. The neighboring vehicles exchange their predicted trajectories at each sample time to predict the conflicts. Then, decentralized coordination and cooperation is performed to resolve the predicted conflicts. The Coordination part consists of online recalculation of the directed interaction graph topology to label conflicting vehicles as leader or follower. Between two conflicting vehicles, the vehicle with higher speed is labeled as leader and the other as follower. The Cooperation part consists of two simple rules, referred to as Heading-rule and Velocity-rule, which are often employed by human pedestrians to avoid potential collisions. The Heading-rule is first employed by both leader and follower to resolve the conflict. If it is not feasible to resolve the conflict by Heading-rule then the Velocity-rule is employed to decelerate the follower and accelerate the leader until the conflict is resolved. Numerous simulations of a team of unicycles are used to illustrate the proposed approach.

Decentralized control of multiple vehicles with limited communication bandwidth

A new algorithm is proposed for decentralized receding horizon control (DRHC) of multiple cooperative vehicles with limited communication bandwidth. The proposed algorithm uses a communication bandwidth allocation approach to optimize the teaming behavior. With the presented DRHC framework each vehicle uses a model of its neighboring vehicles to predict their future intention. The results from stability and performance analysis of such DRHC architecture suggest that the mismatch between actual and predicted plans of neighboring vehicles plays an important role in stability and performance of the team. Hence, the key idea with the proposed algorithm is to reduce the mismatch parameter by means of an efficient communication bandwidth allocation. Simulation results for the formation control of a team of rotorcrafts show the effectiveness of the proposed algorithm.