Towards Effective User-Guided Robot Search

In recent years there has been a growing interest amongst Human-Robot Interaction (HRI) researchers in the usage of robotic systems for exploration or search missions [6, 3, 4]. In the traditional approach to robot supervision by humans, knows as camera guided teleoperation [3], the operator navigates the robot while interpreting camera imagery. Both tasks require constant attention, and contribute to operator’s workload. Automating the robot navigation task, as suggested in [5], is one possible step to reduce workload. Another step is to eliminate the need to view live video, thus turning the interface to use asynchronous video. Recent work [3] on multi-robot interfaces investigated the usage of asynchronous operation schemes in the context of Urban Search And Rescue (USAR) missions [2]. With the asynchronous approach, robots explore the environment autonomously, while the operator is presented with recorded imagery (asynchronous) from the robots’ camera, rather than live imagery (synchronous). Asynchronous interfaces have several advantages. First, scaling up the number of robots is easier, since the operator no longer has to observe live video feeds from multiple robots. This can help to reduce operator’s workload when the number of robots is increased [4]. Moreover, communication constraints can limit the ability to stream live video from the robots and to control them in real-time. This could severely affect the operator’s ability to extract useful information and maintain spatial orientation [1]. With the asynchronous approach however, transmission of recorded