Designed based on the underactuated mechanism, HIT/DLR Prosthetic Hand is a multi-sensory flve-flngered bio- prosthetic hand. Similarly with adult's hand, it is simple constructed and comprises 13 joints. Three motors actuate the thumb, index finger and the other three fingers each. Actuated by a motor, the thumb can move along cone surface, which resembles human thumb and is superior in the appearance. Driven by another motor and transmitted by springs, the mid finger, ring finger and little finger can envelop objects with complex shape. The appearance designation and sensory system are introduced. The grasp experiments are presented in detail. The hand has been greatly improved from HIT-ARhand. It was verified from experimentations, the hand has strong capability of self adaptation, can accomplish precise and power grasp for objects with complex shape.

In the scenario where autonomous underwater vehicles (AUVs) carry out tasks, it is necessary to reliably estimate underwater-moving-target positioning. While cameras often give low-precision visibility in a limited field of view, the forward-looking sonar is still an attractive method for underwater sensing, which is especially effective for long-range tracking. This paper describes an online processing framework based on forward-looking-sonar (FLS) images, and presents a novel tracking approach based on a Gaussian particle filter (GPF) to resolve persistent multiple-target tracking in cluttered environments. First, the character of acoustic-vision images is considered, and methods of median filtering and region-growing segmentation were modified to improve image-processing results. Second, a generalized regression neural network was adopted to evaluate multiple features of target regions, and a representation of feature subsets was created to improve tracking performance. Thus, an adaptive fusion strategy is introduced to integrate feature cues into the observation model, and the complete procedure of underwater target tracking based on GPF is displayed. Results obtained on a real acoustic-vision AUV platform during sea trials are shown and discussed. These showed that the proposed method is feasible and effective in tracking targets in complex underwater environments.

Robot machine capture protein seafood like sea cucumber and seashell is expected to play a great role in food economy improvement and divers protection. In order to realize robot capture, a sea organism absorptive type remotely operated vehicle has been designed with pilot operation and vision based autonomous capture modes. A novel region-based fully convolutional network with deformable convolutional networks has been developed to realize organism target recognition. The comparisons in off-line experiments have verified its advantages. In order to realize organism target following and capture control, a novel learning-based type-II fuzzy controller has been developed. Through online fuzzy rule optimization and learning, the controller can realize organism target following and capture control under image coordinate without vehicle horizontal velocity or position information in complicated submarine environment. Field trials have been made in the Zhangzidao Island of China with the designed absorptive type Remotely Operated Vehicle. The trials manifest the designed absorptive type Remotely Operated Vehicle can realize online organism target recognition, following and capture in the real submarine environment.

To realize underwater accurate vision-based operations, this article proposes a hybrid visual servoing control scheme for underwater vehicle manipulator systems (UVMSs) toward a moving target. First, the position-based visual servoing (PBVS) scheme is purposed based on the binocular camera. Second, to reduce the influence of hand-eye system uncertainties, an uncalibrated visual servoing (UVS) control scheme based on the monocular camera is presented. The partitioned Broyden’s method is utilized to estimate the Jacobian matrix, and a modified adaptive Broyden’s class method is proposed to obtain the approximation of the residual matrix. Afterward, in order to realize long range accurate operations, a novel hybrid visual servoing scheme with Jacobian matrix fusion algorithm is presented not only to combine the PBVS scheme and the UVS scheme but also to avoid the trial movements’ influence. Simulations and an experiment are conducted to testify the robustness and effectiveness of the presented scheme.