Filtering by
- All Subjects: MATLAB
- Creators: Mechanical and Aerospace Engineering Program
- Creators: Electrical Engineering Program
Energy efficient optimal formation control of a multiple quadrotor UAV system with uncertain payload
This thesis presents the design and simulation of an energy efficient controller for a system of three drones transporting a payload in a net. The object ensnared in the net is represented as a mass connected by massless stiff springs to each drone. Both a pole-placement approach and an optimal control approach are used to design a trajectory controller for the system. Results are simulated for a single drone and the three drone system both without and with payload.
The objective goal of this research is to maximize the speed of the end effector of a three link R-R-R mechanical system with constrained torque input control. The project utilizes MATLAB optimization tools to determine the optimal throwing motion of a simulated mechanical system, while mirroring the physical parameters and constraints of a human arm wherever possible. The analysis of this final result determines if the kinetic chain effect is present in the theoretically optimized solution. This is done by comparing it with an intuitively optimized system based on throwing motion derived from the forehand throw in Ultimate frisbee.
The user inputs target position, muzzle position, and estimated environmental parameters to the system. Then, an aim vector would be calculated to hit the target under estimated conditions. Because the eleven trajectory parameters likely cannot all be precisely known, this solution will have some error. In real life, the system would use feedback from real shots of a firearm to correct for this error. For this project, a real-world proxy simulation was created that had built-in random error and variations in the parameters. The correction algorithm uses the error data from all previous shots to calculate adjustments to the original aim vector, so that each successive shot becomes more accurate. The system was tested with specifications of a common rifle platform, with estimated parameters and variations for a location in Tempe, AZ (since data for an urban area is readily available compared to a point in the wilderness). Results from this testing revealed that the system can “hit” a 2-meter-radius circular target in under 30 shots. When the errors and variations in parameters were halved for the real-world stand-in simulation, the system could “hit” a circular target with 0.55 meter radius in less than 25 shots. After analysis, it was found that the corrected aim angles converged on values, suggesting that the correction algorithm functions as intended (taking into account all past shots). Generally, it was found that any reduction of the means and standard deviations of parameter error improved the ability of the system to hit smaller targets, or hit the same target with less shots.