Goal-Oriented Semantic Communication for Multi-Robots Odor Source Localization

In robotic odor source localization (OSL), the chemical plume is released by an odor source and dispersed by airflow. The cooperation of multi-robots can shorten the time to find and track the plume, so as to reduce the time to locate the odor source. Current research works on multi-robots OSL mainly focus on the control process, while the communication process is neglected. Although traditional bit-oriented communication framework can shed light on this issue, it may cause redundant information transmission and a waste of wireless resources. To this end, this paper designs a goal-oriented semantic communication framework (GSCF) for multi-robots OSL. Specifically, for the plume finding stage, we propose an age of information (AoI) based queue ranking mechanism to prioritize fresh uplink sensor data. For the plume tracking stage, we propose effectiveness-aware uplink sensor data retrieval, value of information (VoI) based queue ranking and AoI based sensor data discard mechanisms to guarantee the successful transmission of high-concentration sensor data. We further propose a maximum concentration gradient based gradient fusion (MCG-GF) algorithm to jointly utilize the effectiveness level information of multiple robots. Simulation results demonstrate that our proposed GSCF can reduce the OSL time by at least 11.12% when compared to the traditional bit-oriented multi-robots OSL approach, which verifies that prioritizing semantically critical sensor data can substantially improve localization efficiency.