Relative Pose Based Redundancy Removal: Collaborative RGB-D Data Transmission in Mobile Visual Sensor Networks

In this paper, the Relative Pose based Redundancy Removal (RPRR) scheme is presented, which has been designed for mobile RGB-D sensor networks operating under bandwidthconstrained operational scenarios. Participating sensor nodes detect the redundant visual and depth information to prevent their transmission leading to a significant improvement in wireless channel usage efficiency and power savings. Experimental results show that wireless channel utilization is improved by 250% and battery consumption is halved when the RPRR scheme is used instead of sending the sensor images independently. Note to Practitioners— The invention of low-cost RGB-D cameras has made large-scale, high-resolution 3-D sensing for mapping, immersive telepresence, surveillance, or environmental sensing tasks easily achievable. A network of mobile robots, equipped with the RGB-D cameras can work collaboratively and autonomously to rapidly obtain the 3-D map of an area of interest. Especially in the cases of natural or industrial disasters such as earthquakes, nuclear reactor accidents, or building fires, knowing what is happening in a timely manner can be critically important, and using a network of cooperating camera equipped robots as “mobile visual sensor networks” will surely reduce the mapping time while enhancing the overall reliability of the operation through redundancy. However, the volume of visual and depth data generated by these robots is large, which presents a challenge for efficient data transmission and storage, particularly over the shared wireless channels. The problem is further exacerbated by the operation of the robots in such hostile environments leading to communication difficulties. In this paper, we present a scheme to alleviate this problem. Our scheme allows the robots to detect redundant visual information and prevent its transmission, gathered when multiple cameras have overlapping fields-of-view. The scheme is computationally lightweight, and so can be implemented on battery operated embedded systems without any difficulty. Additionally, we also significantly extend the battery life of the robots by reducing the transmission load.