Robot Identification and Localization with Pointing Gestures

Proximity Human-Robot Interaction Using Pointing Gestures and a Wrist-mounted IMU

“…One key component is relative localization; Gromov et al [1] report extensive experiments validating this component independently, demonstrating its accuracy and robustness, also for robot identification; therefore, in this paper we focus on validating other functionalities. Previously we also shown Interaction using pointing gestures with a slow-moving robot equipped with a laser turret for a real-time visual feedback.…”

“…Many solutions to this problem are adopted in robotics, including fiducial markers [29], GPS [30], optical motion capture systems or ultra-wideband (UWB) localization systems [31,32]. In this paper we use the pointing gesture itself to fix the relative localization, implementing [1].…”

“…We first explain how, relying on the approach proposed in our previous work [1], we determine the transformation T between {R} and {H} assuming that during a time period τ the operator was in fact following a given robot; i.e. we assume triggering and robot identification are known.…”

“…Step 1 simultaneously solves triggering, robot identification and relative localization by extending an approach we recently proposed [1]: such approach (briefly described in Section III-B) compares the arm movements in the last few seconds with the trajectory of the robot in the same time window in order to determine the relative pose of an operator and a robot. In step 2, the robot itself provides a real-time visual feedback about the pointed locations that the system is reconstructing, such that the user has the opportunity to compensate for inaccuracies -similarly to how a laser scope improves accuracy of snipers, and seeing the mouse pointer allows one to interact with small elements on a screen.…”

“…The main contribution of this system paper is the design (Section III), implementation (Section IV) and experimental validation (Section V) of a self-contained pointing-based proximity interaction technique. The system relies on an approach [1] we recently proposed for estimating the relative pose of a user and a mobile robot; in this paper, this approach is used as a component and extended to also solve the interaction triggering task. Moreover, our system implements the idea of using the robot position itself to provide visual feedback about the reconstructed pointed location, which we recently demonstrated in a video abstract [5].…”