As the close perceptual sibling of vision, the sense of touch has historically received less than deserved attention in both human psychology and robotics. In robotics, this may be attributed to at least two reasons. First, it suffers from the vicious cycle of immature sensor technology, which causes industry demand to be low, and then there is even less incentive to make existing sensors in research labs easy to manufacture and marketable. Second, the situation stems from a fear of making contact with the environment, avoided in every way so that visually perceived states do not change before a carefully estimated and ballistically executed physical interaction. Fortunately, the latter viewpoint is starting to change. Work in interactive perception and contact-rich manipulation are on the rise. Good reasons are steering the manipulation and locomotion communities’ attention towards deliberate physical interaction with the environment prior to, during, and after a task. We approach the problem of perception prior to manipulation, using the sense of touch, for the purpose of understanding the surroundings of an autonomous robot. The overwhelming majority of work in perception for manipulation is based on vision. While vision is a fast and global modality, it is insufficient as the sole modality, especially in environments where the ambient light or the objects therein do not lend themselves to vision, such as in darkness, smoky or dusty rooms in search and rescue, underwater, transparent and reflective objects, and retrieving items inside a bag. Even in normal lighting conditions, during a manipulation task, the target object and fingers are usually occluded from view by the gripper. Moreover, vision-based grasp planners, typically trained in simulation, often make errors that cannot be foreseen until contact. As a step towards addressing these problems, we present first a global shape-based feature descriptor for object recognition using non-prehensile tactile probing alone. Then, we investigate in making the tactile modality, local and slow by nature, more efficient for the task by predicting the most cost-effective moves using active exploration. To combine the local and physical advantages of touch and the fast and global advantages of vision, we propose and evaluate a learning-based method for visuotactile integration for grasping.