Given the wide range of robotic applications that can benefit from human understanding and modeling, such as service, rehabilitation and assistive robotics, this topic attracts increasing attention in the robotics society.

On the one hand, it is widely known that accurate navigation is more than necessary for an intelligent system to accomplish its assigned tasks. Hence, as the mobile robot navigates through an unknown environment, it has to construct a map of its surroundings and simultaneously estimate its pose within this map, i.e., perform simultaneous localization and mapping (SLAM). On the other hand, human-robot coexistence adds dynamic features to the navigation task, highly dependent on the individual's activity, intentions and affects, as well as additional requirements with regard to safety and security. Thus, a wide range of techniques and applications have been proposed to address one or more of these challenges, formulating the demanding and interdisciplinary task of socially aware navigation. The task focuses on human modeling, i.e., human pose estimation, human action recognition, language understanding, affective computing, and their efficient integration in the navigation task. As a result, systems that engage in socially aware navigation can excel in adverse, dynamic and highly interactive environments while sustaining lightweight operation capacities.

The main objective of a human-aware navigation pipeline is to facilitate human-robot coexistence in a shared environment. Such coexistence requires the efficient parallel realization of each member’s goals without needless external interceptions or delays and the successful completion of specific common tasks. On top of that, the robotic agent is expected to inspire a sense of trust and friendliness in the human, mainly realized when the agent operates concisely, adaptive, transparent, and naturally. Thus, robot navigation pipelines must employ enhanced human understanding and modeling techniques, capturing those features that mainly affect the efficiency of the task. As a result, it becomes increasingly vital to develop robust, lightweight action and affect estimation solutions based on robotics sensory data and sensory capacities, like active vision and dynamic environments.

Furthermore, apart from the perception layer, enhanced policies for the robotic agent must be developed that exploit the provided high-level human-centered knowledge to generate adaptive navigation paths. Such paths typically trade off execution time efficiency with the level of comfort to meet the corresponding safety and fluidity requirements. Finally, computational efficiency and real-time operation capacities always limit the introduced solutions.

These issues call for new methods to tackle the social-aware navigation problem, including the perception and action spectrum. Therefore, this Research Topic will include articles focusing on techniques that use different sensor modalities to achieve or improve long-term and large-scale social-aware navigation pipelines.
The scope of this Research Topic should include, but are not limited to, the following:
- Active vision for social-aware navigation
- Place recognition in indoor environments with dynamic scenes
- Real-time action recognition in low-energy consumption platforms
- Robust lightweight affect computing and behavioral modeling for robotics
- Adaptive and fluid human-aware robot navigation policies
- Language instructions for robot navigation

Submit your idea

Impact

Frontiers in Robotics and AI

Participating Journals

Impact