Design of model predictive control via learning automata for a single UAV load transportation

Adaptive predictive control of a differential drive robot tuned with reinforcement learning

“…Note that r(n) is dependent on the product of and r b in Equation (16). Therefore, numerically, we are able to fix the value of r b (in our case, to 1) and only tune .…”

“…For example, a reinforcement learning (RL) technique called learning automata (LA) has been used to tune the gains of proportional-integral-derivative (PID) controllers. 13 Recently, the application of this learning technique to linear 14,15 and nonlinear MPC 16 has been explored. The application areas for such works have been unmanned aerial vehicles [13][14][15] and ground vehicles.…”

“…The application areas for such works have been unmanned aerial vehicles [13][14][15] and ground vehicles. 16 While previous work has demonstrated LA can be used to select suitable parameters, no formal analysis of the optimality of the resultant parameters yet exists.…”

“…As it is typically the case for real-time application of MPC on autonomous vehicles, this paper focuses on formulations involving linear plant models and quadratic objective functions. For a similar investigation of this techniques on nonlinear systems, refer to the work of Cabral et al 16 The remainder of this paper is organized as follows: Section 2 introduces the mathematical notation used throughout this paper; Section 3 presents the robot dynamic model, including the method of linearization; Section 4 presents the MPC formulation of the problem including the definition of constraints; Section 5 presents the LA technique and explains how it is applied to the MPC design; Section 6 describes the simulation and experimental environments; Section 7 presents the results of the learning procedure and the experimental validation of the controller; and Section 8 concludes this paper.…”

Nonlinear Control of a Multilink Aerial System and ASEKF-Based Disturbances Compensation

“…It includes high-precision weather monitoring, swarm-based remote sensing, search and rescue, reconnaissance, parcel delivering, homeland security and surveillance, precision agriculture, disaster assessment, and infrastructure inspection [1], [2], [3]. The wide variety of prior works on aerial transportation witnesses the evolution on this field ranging from single [4], [5], [6] to multi-drone configurations [7], [8]. The latter reveals plenty of scientific and technological challenges with enormous potential regarding the industrial sector.…”