In the last couple of years, advancements in the deep learning, especially in convolutional neural networks, proved to be a boon for the image classification and recognition tasks. One of the important practical applications of object detection and image classification can be for security enhancement. If dangerous objects or scenes can be identified automatically, then a lot of accidents can be prevented. For this purpose, in this paper we made use of state-of-the-art implementation of Faster Region-based Convolutional Neural Network (Faster R-CNN) based on the monitoring video of hoisting sites to train a model to detect the dangerous object and the worker. By extracting the locations of them, object-human interactions during hoisting, mainly for changes in their spatial location relationship, can be understood whereby estimating whether the scene is safe or dangerous. Experimental results showed that the pre-trained model achieved good performance with a high mean average precision of 97.66% on object detection and the proposed method fulfilled the goal of dangerous scenes recognition perfectly.

Disability to move hands perfectly is one of the most severe human physical disabilities, and it is mostly common among adults or those who have experienced serious accidents. It is desired to find a methodology to restore the motion of the hand. To this aim, we proposed and evaluated the use of wearable robots (i.e., a smart glove) for clinical therapy that is equipped with shape memory alloy (SMA) actuators. The developed robotic tool, which is a smart glove, uses the structure of human hand as a base mechanism. This glove compensates the weakness of the hand muscles utilizing the forces produced by SMA actuators. The attractive property of high force to weight ratio in SMAs makes this glove a good candidate to be used as a wearable system. The glove actuation mechanism is tendon driven. For every finger, two active Degrees Of Freedom (DOFs) are supported in design. Consequently, four tendons are considered for activating each DOF in order to complete opening and closing phases of the fingers. Totally, twenty tendons are used for rehabilitation of a hand through the glove. Using kinematic relations between tendon length and finger movement, the required deflection of each tendon is extracted. Since a short length of SMA wires cannot provide an enough displacement of tendons in the glove, therefore, an extra mechanism was embedded to the developed glove to support the required length of SMA wires. The SMA actuators were selected and mounted on the system to support the tendons of the mechanism effectively. Moreover, the gripping force provided by the developed glove was also studied. To this end, an analysis was accomplished to extract the relationship between tendon actuation and gripping force of the glove. The obtained results offered a proper model for such a tendon driven glove. Coupling the model of the SMA actuators to that of the tendon driven glove, a composite model of smart glove was extracted. The aforementioned model was simulated numerically. Furthermore, the results were compared with those of the real-world prototype. The obtained results revealed the accuracy of the developed model which will be then employed for both system optimization and model-based control design.

Artificial or “bionic” limbs have been the subject of considerable research, TV shows, and dreams by children. The “Six Million Dollar Man” show was about a man who received artificial limbs after his own were lost in an accident. To get students interested in practical engineering, the current work showcases a simple artificial arm that produces greater force than a typical man, demonstrates the capability of Rubber Muscle Actuators (RMA), and provides a portable “arm wrestling platform” for student recruitment efforts. The actuators for “Kingsville Arm One & Two” are McKibben-like actuators made from fiber-reinforced elastomeric composites. These actuators offer excellent strength-to-weight ratios and contract similar to a human muscle. RMAs produce greater force and have less “blow-outs” than typical McKibben actuators because of optimized braid angles and ends that transfer loads through the braid fibers. Kingsville Arm One (KA1) was developed in just two weeks. It consisted of carbon/fiberglass/epoxy composite tubular bones, a metal clevis “elbow” and four RMAs. With considerable effort, a very large student was able to overcome the force generated in an “arm wrestling” contest. KA1’s actuators had end attachments that transferred loads well and enabled flexibility, but easily tore and had air leaks. Kingsville Arm Two (KA2) had new “bones” and RMAs. Although slightly smaller diameters, the KA2 RMAs produced comparable forces to the KA1 RMAs and had molded end attachments. The rigid ends did not allow as much rotation as expected and necessitated using just 2 RMAs. With only two RMAs, KA2 produced approximately the same “arm strength” as KA1. Future work will focus on flexible but durable RMA molded ends, life-like skins and a realistic “hand.”