Effective multi-agent teamwork can be facilitated by using personas to decompose goals into lower-level team subtasks through a shared understanding of multi-agent tasks. However, traditional methods for role discovery and assignment are not scalable and fail to adapt to dynamic changes in the environment. To solve this problem, we propose a new framework for learning dynamic role discovery and assignment. We first introduce an action encoder to construct a vector representation for each action based on its characteristics, and define and classify roles from a more comprehensive perspective based on both action differences and action contributions. To rationally assign roles to agents, we propose a representation-based role selection policy based on consideration of role differences and reward horizons, which enables agents to play roles dynamically by dynamically assigning agents with similar abilities to play the same role. Agents playing the same role share their learning of the role, and different roles correspond to different action spaces. We also introduce regularizers to increase the differences between roles and stabilize training by preventing agents from changing roles frequently. Role selection and role policy integrate action representations and role differences in a restricted action space, improving learning efficiency. We conducted experiments in the SMAC benchmark and showed that our method enables effective role discovery and assignment, outperforming the baseline on four of the six scenarios, with an average improvement in win rate of 20 $$\%$$% , and is effective in hard and super hard maps. We also conduct ablation experiments to demonstrate the importance of each component in our approach.

Simultaneous Localization and Mapping (SLAM) enables mobile robots to autonomously perform localization and mapping tasks in unknown environments. Despite significant progress achieved by visual SLAM systems in ideal conditions, relying solely on a single robot and point features for mapping in large-scale indoor environments with weak-texture structures can affect mapping efficiency and accuracy. Therefore, this paper proposes a multi-robot collaborative mapping method based on point-line fusion to address this issue. This method is designed for indoor environments with weak-texture structures for localization and mapping. The feature-extraction algorithm, which combines point and line features, supplements the existing environment point feature-extraction method by introducing a line feature-extraction step. This integration ensures the accuracy of visual odometry estimation in scenes with pronounced weak-texture structure features. For relatively large indoor scenes, a scene-recognition-based map-fusion method is proposed in this paper to enhance mapping efficiency. This method relies on visual bag of words to determine overlapping areas in the scene, while also proposing a keyframe-extraction method based on photogrammetry to improve the algorithm’s robustness. By combining the Perspective-3-Point (P3P) algorithm and Bundle Adjustment (BA) algorithm, the relative pose-transformation relationships of multi-robots in overlapping scenes are resolved, and map fusion is performed based on these relative pose relationships. We evaluated our algorithm on public datasets and a mobile robot platform. The experimental results demonstrate that the proposed algorithm exhibits higher robustness and mapping accuracy. It shows significant effectiveness in handling mapping in scenarios with weak texture and structure, as well as in small-scale map fusion.

SLAM (Simultaneous Localization and Mapping) is mainly composed of five parts: sensor data reading, front-end visual odometry, back-end optimization, loopback detection, and map building. And when visual SLAM is estimated by visual odometry only, cumulative drift will inevitably occur. Loopback detection is used in classical visual SLAM, and if loopback is not detected during operation, it is not possible to correct the positional trajectory using loopback. Therefore, to address the cumulative drift problem of visual SLAM, this paper adds Indoor Positioning System (IPS) to the back-end optimization of visual SLAM, and uses the two-label orientation method to estimate the heading angle of the mobile robot as the pose information, and outputs the pose information with position and heading angle. It is also added to the optimization as an absolute constraint. Global constraints are provided for the optimization of the positional trajectory. We conducted experiments on the AUTOLABOR mobile robot, and the experimental results show that the localization accuracy of the SLAM back-end optimization algorithm with fused IPS can be maintained between 0.02 m and 0.03 m, which meets the requirements of indoor localization, and there is no cumulative drift problem when there is no loopback detection, which solves the problem of cumulative drift of the visual SLAM system to some extent.

Energy efficiency and balanced load are essential for the development of clustering algorithms for Mobile Wireless Sensor Networks (MWSNs). Aiming at the problems of uneven load distribution and high overhead of command interaction in existing clustering schemes, we propose an efficient low-energy balanced clustering based on fast Voronoi diagram division (ELBCFVD) algorithm. The innovation of this algorithm is that by judging the information value of the data acquired by the nodes, selecting the representative nodes that satisfy the fast solution of the Voronoi diagram division scheme, optimizing the reconfiguration process of the clusters together with the correspondence of the stable workflows in the mobile scenarios, effectively reducing the overall amount of interaction of the control commands, and realizing the redistribution of the network’s energy consumption and balancing the routing loads of various parts of the network. Detailed simulation analysis shows that the proposed ELBCFVD algorithm significantly reduces the frequency of cluster reconstruction and cluster head rotation, balances the communication load among clusters, and achieves 104.4%, 45.8%, and 15.4% network lifetime extension compared to LEACH-M, EEMCS, and HLBC algorithms, respectively.

Node mobility improves the self-deployment capability of the network; meanwhile, it also leads to frequent interruption of communication links and severe packet loss. Mitigating the negative impact of node movement on cluster stability is a new challenge. Existing clustering protocols try to use multi-hop data transmission, but they do not deal with the increase in communication overhead. This paper proposes a distributed node status judgment-based weighted clustering algorithm to solve the problems of easily broken communication links and excessive node reaffiliation in mobile wireless sensor networks (MWSNs). The protocol establishes intra-cluster second-level communication in order to solve the problem of the sudden interruption of dynamic communication links. A node status judgment algorithm was constructed to analyze the motion behavior of sensor nodes, distinguish the node states, and screen multiple communication nodes, thereby alleviating the transmission delay caused by complex routing. The extended Kalman filter (EKF) was used to filter the sensor noise in a non-ideal environment and to predict the actual position of nodes. The simulation results explain that the proposed protocol can effectively reduce node reaffiliation and the dominant set’s update frequency when the node runs at medium and high speeds while simultaneously maintaining low energy consumption.