This paper addresses the distributed task allocation problem for maximizing the total number of successfully executed tasks of multirobot systems. Due to the deadline time of tasks and fuel limits of robotic vehicles, not all tasks can be successfully executed sometimes. Based on the performance impact (PI) algorithm, an effective and efficient performance impact (EEPI) algorithm is proposed, its novelty lies in its cost function and task release procedure. The fundamental ideas of the proposed cost function are as follows. First, the traveling time from the initial position of each vehicle to the positions of its tasks is minimized, so that more time can be left for the vehicle to execute more tasks due to the limited fuel. Second, the start time of each task should be close enough to its deadline, so that tasks with earlier deadlines can be assigned earlier than those with later deadlines. To avoid invalid removal performance impacts (RPIs) and inclusion performance impacts (IPIs), the tasks assigned to a vehicle are all released if the number of tasks removed by the vehicle during the task removal phase is the most, which further increases the total number of successfully executed tasks. Both simulations and hardware-in-the-loop experiments suggest that compared with the state-of-the-art distributed task allocation algorithms, the proposed EEPI is not only effective in maximizing the number of successfully executed tasks but efficient in saving the number of iterations and time to converge. Note to Practitioners —This work was motivated by the limitations of the existing distributed task allocation algorithms for maximizing the total number of successfully executed tasks. The consensus-based bundle algorithm (CBBA) has been proven to guarantee convergence and 50% optimality under the diminishing marginal gain (DMG) assumption in previously published works. Based on CBBA, a performance impact (PI) algorithm was proposed, and simulations show that it can assign more tasks than CBBA when applied to time-critical scenarios with low task-to-vehicle ratios. Starting from the results of PI, a rescheduling method named PI for maximizing assignments (PI-maxAss) was proposed, which has been demonstrated to assign more tasks than PI with high task-to-vehicle ratios. However, much more iterations and time are required by PI-maxAss to converge to globally consistent assignments because of the rescheduling. Due to the above considerations, an effective and efficient performance impact (EEPI) algorithm is proposed in this paper to maximize the number of successfully executed tasks without any rescheduling. Both simulations and hardware-in-the-loop experiments suggest that compared with the algorithms mentioned above, the proposed EEPI is effective in maximizing the number of successfully executed tasks and efficient in saving the number of iterations and time to converge. In future work, the distributed task allocation problem in which several vehicles execute a task at the same time cooperatively or a vehicle executes several tasks simultaneously will be further addressed.

Remote sensing platforms are often equipped with sensors of multiple spectrums to capture the diverse reflective properties of ground areas, typically including the visible spectrum and the near infrared (NIR) spectrum. Moreover, thermal infrared (TIR) sensors capture the radiated heat of targets and are capable of all-day observation regardless of illumination conditions. By leveraging the complementary features of different spectrums, multi-spectral fusion techniques enhance the precision and robustness of remote sensing object detection methods. In this article, we present an object detection method for remote sensing imagery named Multi-spectral Detection Transformer (Multi-spectral DETR). The model fuses multi-spectral features with deformable attention and utilizes fused features for object detection. The Multi-spectral Deformable Attention (MDA) fusion block integrates the flexibility of dynamic weights with the principle of fusion based on local regions. Then, we propose a simple yet effective oriented object detection scheme based on angle prediction. Finally, we introduce a novel cross-teaching method between single-spectral and multi-spectral models, which alleviates the spectral interference issue caused by inconsistent target visibility. Experimental results demonstrate that Multi-spectral DETR achieves state-of-the-art results on both the RGB-NIR VEDAI and the RGB-TIR DroneVehicle datasets.

This letter presents a novel method to solve the decentralized task allocation problem for large-scale multi-agent systems (MASs), with an emphasis on communication efficiency. Conventional methods typically depend on sharing the localized task allocation plans of various agents across a communication network to facilitate a globally consistent task allocation plan for the entire MAS. However, communication capabilities are limited in most practical scenarios, which may sometimes fail to meet the demands of decentralized task allocation, especially when dealing with large numbers of agents and tasks. To address these challenges, a grouping performance impact (GPI) algorithm is proposed to minimize the total number of communications within the MAS while maintaining high task allocation performance. Agents are partitioned into various groups, with a leader agent utilized in each group to generate task allocation plans for the other follower agents. Additionally, a decentralized task allocation method for agent groups is proposed, incorporating a novel cost scheme designed to maximize the number of successfully performed tasks. Comprehensive simulations demonstrate that the proposed GPI algorithm outperforms the state-of-the-art decentralized methods by reducing communication overhead and improving the total number of successfully performed tasks.