Multi-robot box-pushing in presence of measurement noise

The paper aims at solving a multi-robot box-pushing problem in the presence of noisy sensory data using evolutionary algorithm. In the box-pushing problem by twin robots, the range data obtained by the robots at any instant of time are measurements, and the torque and/or force to be developed by the robots for a local movement of the box are estimators. We here use torques and forces developed by the robots to construct two objectives on minimization of the total energy consumed and the total time required for a local movement of the box in the presence of noisy sensory data. The box-pushing problem is solved using the proposed extended version of the noisy non-dominated sorting bee colony (ENNSBC) algorithm to handle noise in the objective surfaces. Experiments undertaken in both simulation and real-world platforms reveal the superiority of the proposed ENNSBC to other competitor algorithms to solve the box-pushing problem with respect to the performance metrics defined in the literature.