The information was obtained locally and was limited to the position of only a few contact points and a limited range of surface normal at the contact points. They were able to achieve the goal by using simple interpretation tree and pruning mechanisms.Allen and Roberts [17] considered a database consisting of only six objects and presented www.selleckchem.com/products/U0126.html a method for recognition of the objects using tactile image data. The gross shape of each object was first recovered through minimization of an error term. Next, the obtained parameters were utilized in the matching stage to fit the best object from the database to the parameters. Allen and Michelman [16], proposed an approach which recognized the objects�� shapes through a set of exploratory procedures.
The approach was to initially determine gross object shape and then to use a hypothesis-and-test method to generate more detailed information about an object. The sensory Inhibitors,Modulators,Libraries data that they used in their Inhibitors,Modulators,Libraries experiments were in the form of tactile images which were collected by active tactile sensing. Recently, Gorges et al. [19] presented an algorithm to classify the objects directly from the finger position and tactile patterns measured by an anthropomorphic robot hand. The purpose was to perceive the partial Inhibitors,Modulators,Libraries shape of objects by enfolding them successively using the hand. The sensory input data were tactile images and machine learning methods such as self organizing map and Bayes classifiers were used. They were able to recognize objects but not with a high accuracy.
They further employed planar sensor surface which adapted its orientation passively to the object surface and gathered additional information for a better object recognition Inhibitors,Modulators,Libraries [19].In addition to the studies of object recognition, others have also focused on active tactile sensing for local shape identification of objects, mostly on curvature detection. Okamura and Cutkosky [2] proposed an approach to define and Drug_discovery identify surface curvature features using spherical robotic fingertips. They utilized various sensory data such as contact location, surface normal direction, and fingertip center position and concluded that not all of these information are needed for object shape recreation e.g., contact location.An overview of the previous works on the tactile object discrimination indicates that these studies can be categorized into two major classes.
These are works that have investigated the human tactile system and those that have implemented the knowledge on tactile systems to design and develop dexterous robotic hands. Our study belongs to the latter category. MEK162 clinical trial It addresses the neglected issue of tactile information perception for robotic/prosthetic hands in objects�� local shape detection. Most of the earlier works were limited to a bounded number of known objects for recognition.