Search

In this article, the distributed leader-following formation control problem of networked mobile robots is investigated. The desired formation is specified by a reference trajectory generated by the leader and the followers' desired relative positions with respect to the leader. On the one hand, for any security-aware multirobot systems, the values of the leader's position and orientation are generally not allowed to be transmitted via the inter-robot communication in the case of the information leakage of formation caused by the possible eavesdropping. On the other hand, relative orientations, different from relative positions, are typically difficult for mobile robots to measure directly, which makes the reference orientation unknown to all followers. In order to track the reference trajectory and form a desired formation in the absence of the reference orientation, followers are divided into two groups according to whether they are able to directly measure the relative positions with respect to the leader. The topology of the sensing/communication network among multirobot systems is described by a directed graph containing a directed spanning tree. Then, two observer-based control laws are proposed for two groups of followers, respectively, in both of which the unknown tracking errors are properly estimated. It is rigorously proven that the resulting closed-loop multirobot system is globally uniformly asymptotically stable. Finally, the effectiveness of our approach is illustrated by an experiment conducted on networked TurtleBot3 Burger mobile robots.

Visual localization is a crucial component in the application of mobile robot and autonomous driving. Image retrieval is an efficient and effective technique in image-based localization methods. Due to the drastic variability of environmental conditions, e.g., illumination changes, retrieval-based visual localization is severely affected and becomes a challenging problem. In this work, a general architecture is first formulated probabilistically to extract domain-invariant features through multi-domain image translation. Then, a novel gradient-weighted similarity activation mapping loss (Grad-SAM) is incorporated for finer localization with high accuracy. We also propose a new adaptive triplet loss to boost the contrastive learning of the embedding in a self-supervised manner. The final coarse-to-fine image retrieval pipeline is implemented as the sequential combination of models with and without Grad-SAM loss. Extensive experiments have been conducted to validate the effectiveness of the proposed approach on the CMU-Seasons dataset. The strong generalization ability of our approach is verified with the RobotCar dataset using models pre-trained on urban parts of the CMU-Seasons dataset. Our performance is on par with or even outperforms the state-of-the-art image-based localization baselines in medium or high precision, especially under challenging environments with illumination variance, vegetation, and night-time images. Moreover, real-site experiments have been conducted to validate the efficiency and effectiveness of the coarse-to-fine strategy for localization.