Robotic Implementation of Brain-Inspired Multi-Sensory Integration and Decision Making

Abstract

Robot autonomy becomes increasingly prolific with the development of advanced artificially intelligent technologies. Embodied robotics is related to real-time interaction of the robot platforms with the environment. An important aspect of embod-ied robotics is the use of multi-sensory information fusion for robust autonomous decisions and actions. Biologically-inspired approaches to multi-sensory integration (Gestalt formation) gained popularity in embodied robotics due to their flexibility and robustness in noisy and unpredictable practical scenarios. In this work, we discuss dynamic behaviors in multi-robot interaction which emerge as a result of implementing cognitively-motivated multi-sensory integration and action selection. A testbed using mobile quadruped Spot robots is proposed, to evaluate hypotheses on the formation of sequential communication patterns in robot-robot collaboration.