Enhancing model accuracy using Odometry for relative location

This study introduces a model that is called Odometry Aided System, which predicts a neighbor robot pose relative to local body-fixed coordinate system in relative location task. This new model is designed to be tolerant to lossy and bandwidth-limited communication channel. Odometry Aided System uses odometry to adopt various specific models of robots. By setting estimated pose of odometry to zero after each successful communication, odometry works as the buffer to compress velocity information into transformation matrix. With this mechanism, robots can delay transmission until getting ready to or try transmission until success. The data confirmed the effectiveness of predicting a maneuverable neighbor through communicating over a lossy and low-bandwidth channel. It also shows Odometry Aided System keeps the prediction reliable when the distance between robots is increasing. Odometry Aided System describes the relative motion of two robots in a more practical way without adding hardware or system load. It can be considered a new framework to solve the relative location problem.