An Effective Collaborative Localization Method for Multi-Robot System without Sufficient Landmarks

This paper investigates the collaborative localization problem of multi-robot systems in environments where GNSS signals are denied and landmarks are not sufficient, where an effective low-cost multi-robot collaborative localization method is proposed. The robots in the system are equipped with low-cost sensors for odometry measurement, and they observe each other and communicate with their neighbors to share their orientation and mutually observed position information. In this way, the collaborative localization task is achieved at lower communication and sensor costs. Given these information, the multi-robot collaborative localization problems are converted into solving nonlinear least squares optimization problems via Gauss-Newton algorithm. Simulation experiments verify that method proposed is effective, the localization results are expected to be satisfactory though landmarks are not sufficient.