Virtual Nervous Systems for Self-Assembling Robots - A preliminary report

We define the nervous system of a robot as the processing unit responsible for controlling the robot body, together with the links between the processing unit and the sensorimotor hardware of the robot — i.e., the equivalent of the central nervous system in biological organisms. Here, we present autonomous robots that can merge their nervous systems when they physically connect to each other, creating a virtual nervous system (VNS), see Fig. 1. We show that robots with a VNS have capabilities beyond those found in any existing robotic system or biological organism: they can merge into larger bodies with a single brain (i.e., processing unit), split into separate bodies with independent brains, and temporarily acquire sensing and actuating capabilities of specialized peer robots. VNS-based robots can also self-heal by removing or replacing malfunctioning body parts, including the brain. Over the last ten years, we have been developing the basic technologies required to enable VNSs. We have developed robots that can form physical connections with each other and the corresponding control algorithms that allow them to do so autonomously [2, 3, 4, 5, 6]. However, these physically connected robots retained independent control of their own bodies, so that the physically connected ensemble could display only crude levels of coordination [7]. The final step was to give our robots the capability to fuse their robotic nervous systems into a VNS, see Fig.1. Figure 2 shows how VNSs work in a