Contact-reactive grasping of objects with partial shape information

Abstract

Robotic grasping in unstructured environments requires the ability to select grasps for unknown objects and execute them while dealing with uncertainty due to sensor noise or calibration errors. In this work, we propose a simple but robust approach to grasp selection for unknown objects, and a reactive adjustment approach to deal with uncertainty in object location and shape. The grasp selection method uses 3D sensor data directly to determine a ranked set of grasps for objects in a scene, using heuristics based on both the overall shape of the object and its local features. The reactive grasping approach uses tactile feedback from fingertip sensors to execute a compliant robust grasp. We present experimental results to validate our approach by grasping a wide range of unknown objects. Our results show that reactive grasping can correct for a fair amount of uncertainty in the measured position or shape of the objects, and that our grasp selection approach is successful in grasping objects with a variety of shapes.

Improving robot manipulation with data-driven object-centric models of everyday forces

“…Recent research in this domain includes work by ), Hsiao et al (2010, Dollar et al (2010), Prats et al (2009), Platt Jr et al (2011. Similar in spirit to our approach of using previously captured forces, Chan et al (2012) have characterized the forces involved in humanto-human object transfer and then presented guidelines for robot controllers for the same task.…”

Dexterous grasping under shape uncertainty

“…To deal with various uncertainties in the grasping problem, one approach is to use sensory feedback to perform grasp adjustment locally so as to find stable grasps near the original planned grasp [17,19,18]. For instance, [19] proposes a set of simple and efficient heuristics to reactively correct the alignment error of the PR2 gripper. In [18], a sensor-based grasp primitive of Barrett hand is developed to adapt to the variation of the task conditions.…”

“…The reactive correction strategy is designed to alleviate the need for precise hand-object pose information and hence can be more robust to pose and location uncer-tainty. The main problem of these methods is that, to design the corrective strategy, the grasp is usually limited to a predefined set of grasp primitives [18] or only simple hand kinematics is considered [19]. For a more complex dexterous hand with a possibility to execute a large variety of grasps, it becomes more difficult to design such a corrective strategy.…”

Learning robot in-hand manipulation with tactile features

“…Tactile and haptic feedback are especially useful for in-hand manipulation. Apart from providing information on the object's pose and contact normals [11][12][13][14][15][16][17], tactile sensing can provide greater robustness to variations in object properties [18][19][20][21], perturbations [22,23], and sensing errors [10,18]. Tactile sensing can also aid by detecting grasp instabilities and slippage that can occur while manipulating an object [13,20,24,25].…”