Rat robot autonomous border detection based on wearable sensors

Bio-robots, a novel type of robot based on a brain-machine interface, have shown great potential in search and rescue tasks. Current research is focused on the bio-robot itself, such as locomotion, localization and navigation, but lacks interactions with the external environment. In this paper we propose a new system that allows a rat robot to autonomously explore the border of an unknown field out of sight and then obtain the boundary map. We invented a wearable backpack, which is an embedded system with laser-ranging sensors, inertial measurement units and an ultra-wide band (UWB) module, for the rat robot. Based on the wearable system, a classification method for motion states based on the random forest algorithm and a navigation algorithm based on a finite state machine were developed for the autonomous exploration of the border and tested in a locomotion experiment. With the localization and distance data from the UWB module and laser-ranging sensors, we mapped the distribution of the border using the Ramber-Douglas-Peucker algorithm. The results show that this system could effectively navigate the rat robot to explore the field and accurately detect the border. The accuracy of classification reaches 97.86% and the error rate of border detection is 5.90%. This work provides a novel technology that has potential for practical applications such as prospecting for minerals and search tasks.