Online Evolutionary Swarm Algorithm for Self-Tuning Unmanned Flight Control Laws

Ghiglino, Pablo; Forshaw, Jason; Lappas, Vaios

doi:10.2514/1.g000376

Cited by 19 publications

(8 citation statements)

References 20 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For implementation into the control system design, it is assumed that the quadrotor exhibits spherical aerodynamic characteristics (i.e., constant drag coefficient in all directions) and that body lift and other nonlinear aerodynamic effects are negligible. The sum of the forces ( 14) on the vehicle include thrust, drag (16), and gravity. The torques on the vehicle are a function of the force on each motor and the distance of each motor from the center of mass, (17) - (19).…”

Section: Methods 2: Pid Controller Geo-fence Algorithm a Methodologymentioning

confidence: 99%

“…The inner-loop controller uses constant gains, while the outer loop controller uses gain scheduling to adapt to varying wind conditions. The gain schedule table is determined using the Artificial Bee Colony (ABC) genetic optimization method [15], [16]. This method was selected because it is a gradient-free search, immune to problems related to local optima, and the code was available in open source [17].…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Feasibility of varying geo-fence around an unmanned aircraft operation based on vehicle performance and wind

D'Souza

Ishihara

Nikaido³

et al. 2016

2016 IEEE/AIAA 35th Digital Avionics Systems Conference (DASC)

View full text Add to dashboard Cite

Managing trajectory separation of unmanned aircraft is critical to ensuring accessibility, efficiency, and safety in low altitude airspace. The concept of a geo-fence has emerged as a way to manage trajectory separation. A geo-fence consists of distance buffers that enclose individual trajectories to identify a 'keep-in' region and/or enclose areas that identify 'keep-out' regions. The 'keep-in' geo-fence size can be defined as a static number or calculated as a function of vehicle performance characteristics, state of the airspace, weather, and other unforeseen events such as emergency or disaster response. Given that the fleet of Unmanned Aircraft Systems (UAS) operating in low altitude airspace will be numerous and non-homogeneous, calculating a 'keep-in' geo-fence will need to balance operational safety and efficiency. A recently tested UAS Traffic Management (UTM) prototype used a geo-fence size of 30 meters, horizontally and vertically, for every operation submitted. The goal of this work is to determine the feasibility of a generalized, simple algorithm that calculates geo-fence sizes as a function of vehicle performance and potential wind disturbances. The resulting geo-fence size could be smaller or larger because the vehicle performance in the presence of wind is considered, thus leading to trajectory separation that is safe and efficient.In this paper, two simplified methods were developed to determine the feasibility of calculating a geo-fence as a function of vehicle parameters and wind information. The first method calculates the geo-fence using basic vehicle parameters and wind sensor data in a set of algebraic-geometric equations. The second method models a generic PID control system that uses a simplified set of equations of motion for the plant and uses gain scheduling to account for wind disturbances. It was found that the Algebraic-Geometric Geo-fence Algorithm provides geofence sizes of approximately 15 meters horizontally and 5 meters vertically, which is much smaller than the UTM static value of 30 meters. In the PID Controller Geo-fence Algorithm it was found that the geo-fence size is further reduced to less than 5 meters, horizontally and vertically. These results reveal that implementing geo-fence calculations provide UTM with the ability to schedule and separate operations based on geofences that are dynamic to vehicle capability and environment, which is more efficient than using a single static geo-fence.

show abstract

Section: Methods 2: Pid Controller Geo-fence Algorithm a Methodologymentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Feasibility of varying geo-fence around an unmanned aircraft operation based on vehicle performance and wind

D'Souza

Ishihara

Nikaido³

et al. 2016

2016 IEEE/AIAA 35th Digital Avionics Systems Conference (DASC)

View full text Add to dashboard Cite

show abstract

“…At the same time, scouts are employed to look at an entirely different set of solutions and capture solutions where the objective function is improved from the previous solutions. For additional details regarding this method, reference papers by Abachizadeh et al 16 and Ghiglino et al 17 The optimization process is started by randomly selecting the design variables contained in a solution for each solution set (or employed bee). Based on the resulting value of the objective function, the probability of each solution to minimize the objective function is calculated.…”

Section: Optimal Control Framework: Design Variables and Objectivementioning

confidence: 99%

“…The control solution was optimized via the Artificial Bee Colony (ABC) genetic optimization method. 16,17 This method was selected because it is a gradient-free search, immune to problems related to local optima. This characteristic is important because stable muti-rotor control solutions lie in a small solution space and the simulation should not be at risk of not converging to a solution.…”

Section: Introductionmentioning

confidence: 99%

Developing a Generalized Trajectory Modeling Framework for Small UAS Performance in the Presence of Wind

D'Souza

2017

AIAA Information Systems-Aiaa Infotech @ Aerospace

View full text Add to dashboard Cite

In the context of UAS Traffic Management operations, where varied vehicle types and uncertain winds are expected, there is a need for a trajectory simulation that trades a vehicle-centric approach with a systems level approach to model expected vehicle performance with an uncertain operational environment. This paper focuses on the first phase of this trajectory prediction development. The goal is to understand the behavior of this new framework as applied to a vehicle targeting a specific terminal altitude. Specifically, a generalized six degree of freedom trajectory model was built to identify vehicle performance in the presence of wind. Generalization of this model was achieved by reducing the number of simplifying assumptions and using vehicle performance parameters, that were non-specific to the control system, to drive the control solution. The inner loop control strategy behind this algorithm is to find the forces and torques required to guide the vehicle to an intended altitude within vehicle-specific force and attitude constraints in the presence of winds. The control solution was optimized via the Artificial Bee Colony genetic optimization method. It was successfully demonstrated that this framework can utilize vehicle performance parameters, that were non-specific to the control system, to optimize a control solution for targeting an operational altitude. Additionally, vertical and lateral path deviations from the nominal case were observed in the wind case results. This forms the basis for quantifying the spatial and temporal requirements a vehicle will need for a given operation.

show abstract

“…4 This controller is combined with a nonlinear disturbance observer; the effectiveness of which is evaluated through the experimental results. To obtain an optimal path tracking for an AUV, a reinforcement learning algorithm is integrated with two neural networks in Cui et al 5 For AUV flight control, different control algorithms have been and are being developed such as robust H∞ controller, 6,7 LQR, 8,9 PID control, 10,11 µ synthesis, 12 backstepping, 13,14 SMC, [15][16][17] self-tuning, 18,19 adaptive, 20 gain scheduling, 21 model predictive, 22 soft computing, 23,24 hybrid controller, 25 etc. Each of these motion control algorithms has its own advantages and disadvantages.…”

Section: Introductionmentioning

confidence: 99%

Design Boundary Layer Thickness and Switching Gain in SMC Algorithm for AUV Motion Control

2019

View full text Add to dashboard Cite

SummaryDesigning the boundary layer thickness and switching gain in the nonlinear part of sliding mode controller (SMC) is one of the main subjects in SMC design that needs human experience, knowledge on the amplitude of disturbances, and information about the bounds of system uncertainties. In this paper, to reduce the trial-and-error effort by the designer(s) two different fitness functions in the horizontal and vertical planes are presented and a heuristic method is used for their optimization. The optimal switching gain in the proposed approach properly compensates the unmodeled dynamics, model uncertainty, and external disturbances. Chattering phenomenon in control signals and noise measurement effect are reduced by the optimal selection of boundary layer thickness. This proposed method is applied to an autonomous underwater vehicle (AUV) and evaluated through the real-time and cost-effective manner. The execution code is implemented on a single-board computer (SBC) through the xPC Target and is evaluated by the processor-in-the-loop (PIL) test. The results of the PIL test in the two different test cases indicate that the chattering phenomenon and amplitude of control signal applied to the actuators are reduced in comparison with the three conventional approaches in the AUV motion control.

show abstract

Online Evolutionary Swarm Algorithm for Self-Tuning Unmanned Flight Control Laws

Cited by 19 publications

References 20 publications

Feasibility of varying geo-fence around an unmanned aircraft operation based on vehicle performance and wind

Feasibility of varying geo-fence around an unmanned aircraft operation based on vehicle performance and wind

Developing a Generalized Trajectory Modeling Framework for Small UAS Performance in the Presence of Wind

Design Boundary Layer Thickness and Switching Gain in SMC Algorithm for AUV Motion Control

Contact Info

Product

Resources

About