Adaptive Side-by-Side Social Robot Navigation to Approach and Interact with People

Abstract

This paper presents a new framework for how autonomous social robots approach and accompany people in urban environments. The method discussed allows the robot to accompany a person and approach to other one, by adapting its own navigation in anticipation of future interactions with other people or contact with static obstacles. The contributions of the paper are manifold: firstly, we extended the Social Force model and the Anticipative Kinodynamic Planner [1] to the case of an adaptive side-by-side navigation; secondly, we enhance side-by-side navigation with an approaching task and a final positioning that allows the robot to interact with both people; and finally, we use findings from experiments of real-life observations of people walking in pairs to define the parameters of the human-robot interaction in our case of adaptive sideby-side. The method was validated by a large set of simulations; we also conducted real-life experiments with our robot, Tibi, to validate the framework described for the interaction process. In addition, we carried out various surveys and user studies to indicate the social acceptability of the robots performance of the accompanying, approaching and positioning tasks.

Effects of a Social Force Model Reward in Robot Navigation Based on Deep Reinforcement Learning

“…Several works have been developed using the SFM in order to get socially acceptable trajectories for robots in diverse situations like side-by-side or approaching tasks [7,8,20] adding elements to incorporate anisotropy or other goals. In this work the forces are calculated only for the robot and the resultant force is calculated only with the repulsive forces.…”

Robot Navigation to Approach People Using $$G^2$$-Spline Path Planning and Extended Social Force Model

“…These metrics are based on previous studies on humans [19] and the proxemic rules, proposed by Hall [10]. Furthermore, the limits of the interaction distances used were based on a previous work of our institute [9], and a similar version of these performances is included in [17]. The final position of the robot respect to the approached person was evaluated using three types of performance metrics.…”

“…1-line 4 and 12. The description of these areas was included in [17]. Here only change in the A the P c correspondent to the companion person to P a for the approaching person, and also the physical position of the area B changes because here we have only the approached person and this area is defined by the final goal of the best path.…”

“…Here only change in the A the P c correspondent to the companion person to P a for the approaching person, and also the physical position of the area B changes because here we have only the approached person and this area is defined by the final goal of the best path. Furthermore, the formulation of how the performance was extracted from these areas and the description of the robot area is detailed in [17], next we only explain the general idea to understand it. The Area metric has the maximum performance of 1 when the robot is in the area described by B, since it is the best position to approach the human.…”

“…The model includes a framework to calculate a moving goal taken into account the movement of the approached person, the movement of the group and the best path to go throw the obstacles of the environment. After that in [17] we enhanced the previous approach by computing the best encounter point using a gradient descent method, taking into account all people's predictions. In the encounter point, the robot performs a triangle formation to achieve an engagement with both people.…”