Human-object interaction (HOI) detection seeks complicated relationships between humans and objects, yet struggles persist in correctly associating multiple objects with a single human in complex interaction scenarios. In this paper, we tackle such an issue by introducing a novel interaction anchor that employs flexible strategies across different decoder layers with Barlow constraint and Interactivity-Instance Fusion. The proposed modules are both additive and easily implementable in existing approaches, offering computational efficiency within transformer-based models to compact cross-interactivity. Extensive experiments validate the effectiveness of our method, demonstrating comparable performance on HICO-DET and V-COCO for HOI detection.

We propose a simple approach which, given distributed computing resources, can nearly achieve the accuracy of k-NN prediction, while matching (or improving) the faster prediction time of 1-NN. The approach consists of aggregating denoised 1-NN predictors over a small number of distributed subsamples. We show, both theoretically and experimentally, that small subsample sizes suffice to attain similar performance as k-NN, without sacrificing the computational efficiency of 1-NN.

Following the recent advances in drone technologies, various algorithmic optimization problems related to the effective operation of drones are drawing lots of attentions. This paper considers two interesting multiple-drone-assisted search-and-reconnaissance scenarios, in each of which, the trajectory optimization of multiple drones is of great significance to minimize the latency in the system. In the first scenario, multiple drones, whose moments of mobilization are not necessarily the same, are trying to urgently collect intelligence from a given point of interest, and we would like to minimize the task completion time, i.e., the time period between the moment that the first drone commences its operation to the moment that the intelligence from all of the points are collected, by optimizing their trajectories. In the second scenario, multiple drones with different speeds, are hovering around the same routes to regularly collect intelligence from highly geographically-diversified points of interest over an extended time period, and we would like to minimize the worst-case data refreshment rate, the largest time gap between two consecutive observations over the same point of interest. In this paper, we formally define each problem, prove its NP-hardness, and propose an approximation algorithm for it. We also conduct a simulation to study the performance of our result.

This paper investigates two new groups of trajectory optimization problems which stem from networked multi-robotic systems. In particular, we study how to efficiently collect data from stationary sensor nodes using multiple robotic vehicles such as data ferries under different circumstance. The first group includes two new problems which aim to find the tours and the paths, respectively, of k robot vehicles with different mobilization conditions to collect data from ground sensor nodes with minimum latency. The second group consists of one new problem whose goal is to determine the quality tours of k robot vehicles with different speeds, where each of which follows its corresponding tour to repeatedly collect data from stationary sensors. We prove the three problems are NP-hard and propose constant factor approximation strategies for them. Through a simulation, an analytical study is conducted to evaluate the average performance of our core contribution.