The ability to identify and temporally segment fine-grained actions in motion capture sequences is crucial for applications in human movement analysis. Motion capture is typically performed with optical or inertial measurement systems, which encode human movement as a time series of human joint locations and orientations or their higher-order representations. State-of-the-art action segmentation approaches use multiple stages of temporal convolutions. The main idea is to generate an initial prediction with several layers of temporal convolutions and refine these predictions over multiple stages, also with temporal convolutions. Although these approaches capture long-term temporal patterns, the initial predictions do not adequately consider the spatial hierarchy among the human joints. To address this limitation, we recently introduced multi-stage spatial-temporal graph convolutional neural networks (MS-GCN). Our framework replaces the initial stage of temporal convolutions with spatial graph convolutions and dilated temporal convolutions, which better exploit the spatial configuration of the joints and their long-term temporal dynamics. Our framework was compared to four strong baselines on five tasks. Experimental results demonstrate that our framework is a strong baseline for skeleton-based action segmentation.
Visible light positioning has the potential to be a cost-effective technology for accurate indoor positioning. However, existing approaches often require large amounts of incoming data, usually in the form of high resolution images or dense lighting distributions. Additionally, a line of sight between transmitter and receiver is generally required at all times. In this work, we present a positioning approach that combines measurements from a camera, encoders and a gyroscope. We compare multiple algorithms for fusing these data, namely an extended Kalman filter, a particle filter and a hybrid approach. The end result is a system that provides location estimates even with sparse lighting distributions and temporary outages, yet achieves an average accuracy of 2 to 4 cm. Even in the 95th percentile of the cumulative error distribution, accuracy can be as low as 2 cm and is often lower than 10 cm. Moreover, due to the use of a low-resolution camera (640x480 pixels) and efficient fusion algorithms, the latency is relatively low on a standard laptop (between 5.6 and 21 milliseconds). Even on a low-cost embedded board, latency generally does not exceed 100 milliseconds. We validate our approach experimentally and show that it is robust under a wide range of illumination conditions.
Augmenting the automation level of the inland waterway cargo transport sector, coupled with mechatronic innovation in this sector, could increase its competitiveness. This increase might potentially induce a sustainable paradigm shift in the road-dominated inland cargo transport sector. A key enabler of this envisaged shift may be an inland shore control centre (I-SCC) capable of remotely monitoring and controlling inland vessels. Accordingly, this study investigated the concept and design requirements to achieve an inland I-SCC that provides interaction services when supervising an unmanned surface vessel (USV). This I-SCC can help its operator to develop situational awareness and sensemaking. The conducted experiments offered insights into the performance of both the I-SCC system and its operator, and unlock research on the impact on ship sense and harmony when remotely controlling a USV. The Hull-To-Hull project extends the current I-SCC by providing enhanced motion control. This enhancement enables further performance insights and might improve the future monitoring of USVs. The successful I-SCC construction, the preliminary experiments, and the design-extension demonstrate that the I-SCC can serve as an experimental platform for both mechatronic innovation and human-automation integration research in the inland waterway sector, whilst additionally providing fruitful knowledge for adjacent research domains.
Most indoor positioning systems require calibration before use. Fingerprinting requires the construction of a signal strength map, while ranging systems need the coordinates of the beacons. Calibration approaches exist for positioning systems that use Wi-Fi, radio frequency identification or ultrawideband. However, few examples are available for the calibration of visible light positioning systems. Most works focused on obtaining the channel model parameters or performed a calibration based on known receiver locations. In this paper, we describe an improved procedure that uses a mobile robot for data collection and is able to obtain a map of the environment with the beacon locations and their identities. Compared to previous work, the error is almost halved. Additionally, this approach does not require prior knowledge of the number of light sources or the receiver location. We demonstrate that the system performs well under a wide range of lighting conditions and investigate the influence of parameters such as the robot trajectory, camera resolution and field of view. Finally, we also close the loop between calibration and positioning and show that our approach has similar or better accuracy than manual calibration.