Search

Though adding manipulator arms gives a two-wheeled self-balancing assistant robot (TSAR) the ability of fetching object, the degenerated balancing performance would occur without considering the effect of robot center-of-gravity (CoG) position when the manipulator arms operate. To tackle this problem, a double-loop fuzzy motion control (DFMC) system is proposed to control the TSAR moving, turning, and reaching a desired position while keeping TSAR balanced. A CoG supervising controller is proposed to control the body pitch angle of TSAR with respect to manipulator arms operate. To show the effectiveness of the CoG supervising controller, three engagement scenarios are applied in the presence of external disturbances and system uncertainties. The experimental results show that the proposed DFMC system with considering CoG supervising controller can achieve better motion performance than without considering that. Further, a visual serving technique is used for doing object recognition and a fuzzy guidance control (FGC) is proposed to let the TSAR can implement the object tracking mission. The experimental results show that not only the DFMC system can maintain TSAR balance but also the FGC system can lead TSAR to fetch target object successfully.

This study proposes the multiobjective H2/H∞fuzzy control design for a nonlinear stochastic chaotic system via concurrently optimizing H2and H∞performance indices in a Pareto optimal sense. Using the Takagi–Sugeno fuzzy model to approximate the nonlinear stochastic chaotic system, the multiobjective H2/H∞fuzzy control design problem can be transformed into a linear matrix inequalities (LMIs)-constrained multiobjective optimization problem (an LMIs-constrained MOP). By the help of the LMIs-constrained multiobjective evolution algorithm (LMIs-constrained MOEA), one can obtain the Pareto optimal controller. However, the existing LMIs-constrained MOEA usually couples with a heavy computational load. This study proposes the front-squeezing LMIs-constrained MOEA to resolve such a computational cost problem. Finally, a simulation example is presented to verify the effectiveness of the proposed theories.