For robots, the ideal environment is one that’s orderly and predictable. Unfortunately for robots, the world is anything but. That means robots of all sorts, from autonomous cars to little Roombas, need a way to navigate dynamic landscapes — without crashing into everything.
A new system, developed by PAL Robotics in Spain, offers another possible solution to this problem. Its StockBot, a robot designed to take on the tedious task of retail inventory, can find its way around thanks to radio beacons set up around the room. Those beacons communicate with the robot, and the robot uses the distance between it and the beacons to determine its location.
PAL Robotics recently made the navigation’s system’s code open-source and available on GitHub, and it’s based on ROS (Robot Operating System). ROS already features a system that helps robots figure out where they are, called Adaptive Monte Carlo Localization. Basically, robots rely on AMCL and probability to pin down location.
Probability isn’t the only option for robot localization. Roombas and autonomous vehicles are equipped with cameras and LIDAR sensors to help place themselves on a map. The new range-based localization system developed by PAL Robotics can be used alone, but PAL Robotics sees it being used in tandem with AMCL, to help refine the localization, and to help the robot get out of tricky situations.
Watch the StockBot navigate a room in the video below, and figure out how to get out of a wrong position, all thanks to range-only localization.