This study addresses integrated bus maintenance and vehicle scheduling in a day-to-day dynamic operation setting to ensure that regular preventative maintenance and newly identified predictive maintenance needs are handled promptly. The problem is formulated as a mixed-integer linear programming aiming to minimize operational costs and risks over a day-to-day rolling planning horizon. A two-stage decomposition (TSD) method is proposed to solve this challenging problem. In the first stage, bus maintenance is scheduled on a daily basis. The second stage then optimizes specific bus maintenance and vehicle schedules within each day. With such a problem decomposition, the computational complexity is significantly reduced as the within-day problems become independent. Computational experiments conducted on a real-world bus line demonstrate the effectiveness and superiority of the TSD method. Compared to a one-step solution method and an algorithm in a similar existing study, the proposed TSD method consumes far less computation time, while delivering high-quality solutions. Moreover, the TSD solution method can be further greatly accelerated if the within-day problems of the second stage are solved in parallel with different computers. The computational results also highlight the benefits of the proposed two-stage optimization approach in enhancing operational cost-efficiency and bus resource utilization.

The increasing popularity of earable devices stimulates great academic interest to design novel head gesture-based interaction technologies. But existing works simply consider it as a singular activity recognition problem. This is not in line with practice since users may have different body movements such as walking and jogging along with head gestures. It is also beneficial to recognize body movements during human-device interaction since it provides useful context information. As a result, it is significant to recognize such composite activities in which actions of different body parts happen simultaneously. In this paper, we propose a system called CHAR to recognize composite head-body activities with a single IMU sensor. The key idea of our solution is to make use of the inter-correlation of different activities and design a multi-task learning network to extract shared and specific representations. We implement a real-time prototype and conduct extensive experiments to evaluate it. The results show that CHAR can recognize 60 kinds of composite activities (12 head gestures and 5 body movements) with high accuracies of 89.7% and 85.1% in sufficient data and insufficient data cases, respectively.

Recent research studies have made significant progress in acoustic-based gesture recognition. However, existing methods lack the capability to expand to customized gestures and adapt to different practical environments. We propose a highly scalable gesture recognition system called EchoGest which integrates a well-designed feature-wise transformation layer into prototypical network framework, and accomplishes unseen gesture recognition with a device’s built-in speaker and microphone. Our key insight involves gauging the similarity between query sample representations and class prototypes in the embedding space, and thus enabling the scalability to unseen gestures. Meanwhile, we introduce a feature transformation layer to linearly adjust feature maps and propose an efficient two-stage training strategy to obtain regularized parameters for this layer. Specifically, this layer employs affine transformation to enhance intermediate feature activations and yield more diverse feature distributions for cross-domain recognition, and it improves recognition accuracy by 10% in 1-shot cases. We train the system with a collected a letter gestures (i.e., writing ’A’ to ’Z’) dataset and test it on a digit gestures (i.e., writing ’0’ to ’9’) dataset with 10 volunteers. The results show that EchoGest can recognize unseen digit gestures with an accuracy of 93.7% in 2-shot cases, and 93.2% in the leave-one-user-out testing setting. We also explore a semi-supervised clustering approach in which each user’s data can be used to update his or her prototypes for personalized customization. The comprehensive experiments also verify that EchoGest remain good performance across various environments, age groups, and different devices.

Acoustic-based human gesture recognition (HGR) applications have drawn increasing academic attention in order to overcome the shortcomings of conventional interaction methods on tiny devices. Existing techniques following a learning-based routine requires collecting massive application-specific training data. What is worse, the cross-domain problem induces additional retraining overhead to enable the systems recognize unseen gestures in different environments. This obviously decreases their scalability and prevent them from real-world deployment. Although some recent works propose different few-shot learning solutions to deal with the cross-domain problem in HGR, they possess shortcomings of being application-specific, high training overhead, and/or incapability to recognize unseen gestures. In this paper, we propose PreGesNet, a few-shot acoustic gesture recognition framework based on task-adaptive pretrained networks whose novelty lies in three aspects: i) leveraging pretrained feature extractor which captures generic knowledge of our collected and open-source large-scale gesture datasets; ii) designing task-specific parameter adaptation mechanism to efficiently update the feature extractor to adapt the pretrained feature extractor to each target task; iii) discovering suitable distance metric and task generation strategy which fit HGR application. According to the experiments, when the model is trained with 10 digit gestures, its recognition accuracies of 26 kinds of letter gestures and 8 kinds of other hand gestures can be up to 80.5% and 93.4% with only two shots, respectively. In addition, the average recognition latency of PreGesNet is less than 0.4 second.