In this paper, we use data from the Microsoft Kinect sensor that processes the captured image of a person using and extracting the joints information on every frame. Then, we propose the creation of an image derived from all the sequential frames of a gesture the movement, which facilitates training in a convolutional neural network. We trained a CNN using two strategies: combined training and individual training. The strategies were experimented in the convolutional neural network (CNN) using the MSRC-12 dataset, obtaining an accuracy rate of 86.67% in combined training and 90.78% of accuracy rate in the individual training. Then, the trained neural network was used to classify data obtained from Kinect with a person, obtaining an accuracy rate of 72.08% in combined training and 81.25% in individualized training. Finally, we use the system to send commands to a mobile robot in order to control it.
This article describes the use of the Deep Deterministic Policy Gradient network, a deep reinforcement learning algorithm, for mobile robot navigation. The neural network structure has as inputs laser range findings, angular and linear velocities of the robot, and position and orientation of the mobile robot with respect to a goal position. The outputs of the network will be the angular and linear velocities used as control signals for the robot. The experiments demonstrated that deep reinforcement learning’s techniques that uses continuous actions, are efficient for decision-making in a mobile robot. Nevertheless, the design of the reward functions constitutes an important issue in the performance of deep reinforcement learning algorithms. In order to show the performance of the Deep Reinforcement Learning algorithm, we have applied successfully the proposed architecture in simulated environments and in experiments with a real robot.
The LD (Linz Donawitz) steelmaking process is the most used in the steel industry due to its highvolume capacity and low cost per ton of steel produced. However, the basic oxygen steelmaking process in LD converters is subjected to potential steel charge overflows, often called 'slopping'. Besides yield losses, slopping events can damage the environment and expose employees to danger. More than ever, steelmaking plants need to avoid this type of event to keep producing as environmental impacts are no more tolerable by society. Steelmaking plants already use different methods to monitor and detect slopping events, but they are often limited and unreliable. Therefore, this paper proposes a multi-sensor data fusion process to generate a reliable slopping index to warn operators of potential slopping events and detect the triggered ones. The work is based on sound and image data (67 heats with 27 slopping events) collected on previous trials at a 350-ton LD converter. The Kalman filter was applied as a data fusion agent of two indexes, one resulted from computer vision analysis of the LD converter mouth (image data), the other resulted from digital signal analysis of sound captured on the converter's hood (sound data). Fuzzy sets were applied for adaptative tuning of the Kalman filter to improve the data fusion process. Besides the increase of alarm accuracy and heat classification, the data fusion index worked better on different scenarios and produced a more reliable indicator for a slopping prevention system.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.