Approaching people

Wizard of Oz Interface

How the robot moves, including acceleration to target, deceleration on arrival, and velocity during travel can impact the perception of the robot by those around it. While there is substantial prior work in this area including our own it is often contradictory. Direction of approach for example is often confounded with the the attitude of the person (whether they are sitting, standing or moving). We need both comprehensive in-lab experiments to explore approaches toward a variety of singletons and groups satisfying all three conditions, and present results together. This is a key area where lab results are questionable, and in-the-wild studies across a range of tasks is necessary. Further, we propose to investigate the role of smoothing in the movement of the robot. For example, in the elevator scenario, it may be important to make one movement (when new people enter the elevator), and then remain still, rather than make continual, small adjustments in a confined space. Intrinsic Evaluation Our own work has focused on the evaluation of robot approach movement and it’s impact on HRI task completion; do certain movements result in higher task completion percentages? We find that they do, and we will continue these evaluations in conjunction with the cues already presented. For the acceleration, deceleration and smoothing elements, we will build on the work of Ferland and Tapus to determine the impact on humans around the robot in differing situations and locations. We can also replicate lab-based approach models to determine which conditions impact approach direction and speed, substantially more comprehensively than has been presented before. Human feedback on approach direction, speed, and factors such as acceleration, deceleration and any uncertainty around small adjustments in position will be recorded.

Nick Webb
Associate Professor of Computer Science / Director of Data Analytics

My research interests include Natural Language Processing, Social Robotics and Data Analytics.