Development of Optimized Advanced Control Laws for Trajectory Tracking of Quadcopter

Abstract

Unmanned aerial vehicles (UAVs), particularly quadcopters, have witnessed remarkable advancements, showcasing their capabilities in diverse tasks and object manipulation. These advancements are attributed to sophisticated control systems, novel actuation, sensing technologies, and a deep understanding of aerodynamics and gyroscopic moments. This thesis contributes to the evolution of optimized advanced control laws for trajectory-tracking quadcopters, extending beyond the scope of the initial research. Initially, a nonlinear model of the quadcopter is developed using Lagrange formalism, capturing gyroscopic moments and aerodynamic effects for a comprehensive representation of dynamics. The research introduces innovative optimized control laws, including the Conditioned Adaptive Barrier Function Integral Terminal Sliding Mode Control (CABFIT-SMC), Conditioned Adaptive Barrier-Based Double Integral Super Twisting SMC (CABDIST-SMC), Barrier Function Double Integral SMC, Barrier Function Integral SMC, and Barrier Function-Based SMC, addressing various trajectory-tracking aspects. Lyapunov stability analysis confirms the asymptotic stability of the quadcopter under these control laws. 7 optimization algorithms (i.e., Artificial Bee Colony, Ant Colony, Improved Grey Wolf, Genetic Algorithm, Particle Swarm, Quantum Particle Swarm, and Redfox) are used to optimize the proposed control law and their performances are compared to determine which algorithm yielded the optimal control performance. Furthermore, deep learning and reinforcment learning based techniques are implemented on the nonlinear control laws to explore another realm. The novel introduction of Bidirectional Long Short-Term Memory (BILSTM) networks is explored to address the computational complexity of the Complex Adaptive Barrier Function Integral Sliding Mode Controller (CABFIT-SMC). Training BI-LSTM on CABFIT-SMC performance data offers a cost-effective alternative with significantly reduced computational expenses. A comparative analysis involving Reinforcement Learning (RL) and BI-LSTM is presented, demonstrating BI-LSTM’s ability to emulate CABFIT-SMC trajectories at a fraction of the computational cost. BI-LSTM-CABFIT-SMC demonstrates a more rapid and accurate response, minimizing deviations from the desired trajectory. In a comparative analysis involving Reinforcement Learning (RL) and BI-LSTM, the results showcase the remarkable efficiency o f BI-LSTM. S pecifically, BI -LSTM-CABFITSMC ac hieves a si mulation ti me of 0.0325 s with a computational cost of 22.1752, outperforming RL-CABFITSMC (2.3125 s, 178.6589) and the original CABFIT-SMC (2.21 s, 1255.2166). These results underscore the superior computational efficiency and accuracy of BI-LSTM, positioning it as a cost-effective alternative that emulates CABFIT-SMC trajectories with unprecedented precision. The comprehensive evaluation and validation processes, supported by numerical results, instill confidence in the effectiveness of the proposed control laws. The performance indices, including mean absolute percentage error, root mean square error, integral square error, integral absolute error, integral time absolute error, and integral time square error, further underscore the superior performance of the proposed CABFIT-SMC and BI-LSTM-based controllers. This research lays the foundation for continued innovation in UAV technology, offering efficient solutions to complex computational challenges. This research work provides a robust framework for the development of optimized advanced control laws. The incorporation of BI-LSTM demonstrates a breakthrough in addressing computational challenges, offering cost-effective alternatives with minimal deviation during real-time implementation. These advancements reinforce the suitability of quadcopters for precise trajectory tracking in both civilian and military applications. The comprehensive evaluation and validation processes instill confidence in the effectiveness of the proposed control laws, opening avenues for further innovations in UAV technology.