Leveraging Memory and Attention in a Kinematically Aware Robot: An Ideomotor-Inspired Approach to Implicit Command Understanding from IMU Sensor Data*

As human-robot interaction (HRI) advances, the nuanced interpretation of implicit commands embedded in human gestures becomes paramount for fostering seamless collaboration. In this context, we present a novel machine learning algorithm designed to endow robots with the ability to decipher implicit commands from Inertial Measurement Unit (IMU) sensor data worn at specific locations on the human body. Our approach integrates memory and attention mechanisms inspired by ideomotor cues, allowing the robot to comprehend both temporal and spatial relationships within the sensor data. The attention mechanism operates bidirectionally, enhancing the system’s awareness of the temporal sequence of human movements and the spatial interdependencies between sensor data across different body locations. This unique spatial attention enables the robot to understand the kinematic chain between joints during human motion, accommodating variations in sensor data arising from factors such as height differences and motion range capacity. Drawing on prior research in attention mechanisms, ideomotor cues, and memory augmentation, our algorithm represents a significant advancement in addressing the challenges of implicit command understanding in HRI. The proposed system’s adaptability and nuanced comprehension of human gestures make it wellsuited for diverse anatomies and movement patterns. Through comprehensive experiments, we demonstrate the effectiveness of our algorithm, paving the way for more intuitive and adaptable robotic systems in real-world applications.