Apriltag-Based Localization and Tracking System of Aerial and Ground Unmanned Collaborative Platform

This paper introduces a collaborative tracking and localization framework utilizing aerial and ground platform within indoor environments. The proposed method employs Apriltag recognition to accurately ascertain the poses of the agents. In uncharted settings, a partitioned pathfinding algorithm is implemented, complemented by an obstacle avoidance mechanism leveraging elastic tracking. Based on feasible preliminary trajectories, the approach utilizes a gradient descent algorithm to formulate spatiotemporally optimal tracking paths. Ultimately, an extended Kalman filter integrates kinematics datas, thereby enhancing localization precision. Validation through simulation and real-world experiments substantiates the viability of the algorithm, demonstrating its capability in tracking dynamic targets with localization accuracy and robust stability.