Mobile ad-hoc communication in machine swarms for relative positioning based on GNSS-raw data exchange

The increasing automation of mobile working machines and the progressive use of more than one machine up to machine swarms for cooperative tasks demands information about the relative position between the machines as well as between the machines and their attachments. This is especially necessary and important when carrying out tasks on distributed machines in a formation. Examples for application are parallel harvesting process in agricultural business or the cooperative search for survivors after a snow slide or e.g. an earthquake in urban scenarios. Moreover, it is very important to ensure relative position information in case of partial failure or a poor reception of the GNSS receiver, for example to avoid the collision between the machines. Options to handle this problem are the coupling of Global Navigation Satellite System (GNSS) measurements with measurements of an Inertial Measurement Unit (IMU). A further improvement or stabilization can be done by vision based systems, like 2D- or 3D-camera system using methods of optical flow for motion estimation. Another possibility for improvement for determining the swarm geometry, which will be described in this paper, is the so called swarm positioning method. This method is based on the exchange of the measured GNSS raw data, i.e. range measurements, between each participant in the swarm using a mobile ad-hoc network. Additionally, GNSS raw measurements and inertial measurements are coupled using multiple filters in order to detect degraded GNSS measurements and exclude these from further data processing. The challenge of the mobile ad-hoc network is the time variant network structure and the small available transmission rate in combination with a high demand for quick data exchange. The requirement to respond very flexibly to changes in topology and to compensate the loss of individual network participants as well as to spontaneously integrate further participants forbids the use of a fixed coordinator. Therefore different routing algorithms have to be combined and developed to ensure an information exchange in various scenarios.