Despite decades of research robotic manipulation systems outside of highly-structured industrial applications are still far from ubiquitous. Perhaps particularly curious is the fact that there appears to be a large divide between the theoretical grasp modeling literature and the practical manipulation community. Specifically, it appears that the most successful approaches to tasks such as pick-and-place or grasping in clutter are those that have opted for simple grippers or even suction systems instead of dexterous multi-fingered platforms. We argue that the reason for the success of these simple manipulation systemsis what we call passive stability: passive phenomena due to nonbackdrivable joints or underactuation allow for robust grasping without complex sensor feedback or controller design. While these effects are being leveraged to great effect, it appears the practical manipulation community lacks the tools to analyze them. In fact, we argue that the traditional grasp modeling theory assumes a complexity that most robotic hands do not possess and is therefore of limited applicability to the robotic hands commonly used today. We discuss these limitations of the existing grasp modeling literature and setout to develop our own tools for the analysis of passive effects in robotic grasping. We show that problems of this kind are difficult to solve due to the non-convexity of the Maximum Dissipation Principle (MDP), which is part of the Coulomb friction law. We show that for planar grasps the MDP can be decomposed into a number of piecewise convex problems, which can be solved for efficiently. Despite decades of research robotic manipulation systems outside of highlystructured industrial applications are still far from ubiquitous. Perhaps particularly curious is the fact that there appears to be a large divide between the theoretical grasp modeling literature and the practical manipulation community. Specifically, it appears that the most successful approaches to tasks such as pick-and-place or grasping in clutter are those that have opted for simple grippers or even suction systems instead of dexterous multi-fingered platforms. We argue that the reason for the success of these simple manipulation systemsis what we call passive stability: passive phenomena due to nonbackdrivable joints or underactuation allow for robust grasping without complex sensor feedback or controller design. While these effects are being leveraged to great effect, it appears the practical manipulation community lacks the tools to analyze them. In fact, we argue that the traditional grasp modeling theory assumes a complexity that most robotic hands do not possess and is therefore of limited applicability to the robotic hands commonly used today. We discuss these limitations of the existing grasp modeling literature and setout to develop our own tools for the analysis of passive effects in robotic grasping. We show that problems of this kind are difficult to solve due to the non-convexity of the Maximum Dissipation Principle (MDP), which is part of the Coulomb friction law. We show that for planar grasps the MDP can be decomposed into a number of piecewise convex problems, which can be solved for efficiently. We show that the number of these piecewise convex problems is quadratic in the number of contacts and develop a polynomial time algorithm for their enumeration. Thus, we present the first polynomial runtime algorithm for the determination of passive stability of planar grasps. For the spacial case we present the first grasp model that captures passive effects due to nonbackdrivable actuators and underactuation. Formulating the grasp model as a Mixed Integer Program we illustrate that a consequence of omitting the maximum dissipation principle from this formulation is the introduction of solutions that violate energy conservation laws and are thus unphysical. We propose a physically motivated iterative scheme to mitigate this effect and thus provide the first algorithm that allows for the determination of passive stability for spacial grasps with both fully actuated and underactuated robotic hands. We verify the accuracy of our predictions with experimental data and illustrate practical applications of our algorithm. We build upon this work and describe a convex relaxation of the Coulombfriction law and a successive hierarchical tightening approach that allows us to find solutions to the exact problem including the maximum dissipation principle. It is the first grasp stability method that allows for the efficient solution of the passive stability problem to arbitrary accuracy. The generality of our grasp model allows us to solve a wide variety of problems such as the computation of optimal actuator commands. This makes our framework a valuable tool for practical manipulation applications. Our work is relevant beyond robotic manipulation as it applies to the stability of any assembly of rigid bodies with frictional contacts, unilateral constraints and externally applied wrenches. Finally, we argue that with the advent of data-driven methods as well as theemergence of a new generation of highly sensorized hands there are opportunities for the application of the traditional grasp modeling theory to fields such as robotic in-hand manipulation through model-free reinforcement learning. We present a method that applies traditional grasp models to maintain quasi-static stability throughout a nominally model-free reinforcement learning task. We suggest that such methods can potentially reduce the sample complexity of reinforcement learning for in-hand manipulation.We show that the number of these piecewise convex problems is quadratic in the number of contacts and develop a polynomial time algorithm for their enumeration. Thus, we present the first polynomial runtime algorithm for the determination of passive stability of planar grasps.

Robotic grasping and manipulation remains an active area of research despite significant progress over the past decades. Many existing solutions still struggle to robustly handle difficult situations that a robot might encounter even in non-contrived settings.For example, grasping systems struggle when the object is not centrally located in the robot's workspace. Also, grasping in dynamic environments presents a unique set of challenges. A stable and feasible grasp can become infeasible as the object moves; this problem becomes pronounced when there are obstacles in the scene. This research is inspired by the observation that object-manipulation tasks like grasping, pick-and-place or insertion require different forms of awareness. These include reachability awareness -- being aware of regions that can be reached without self-collision or collision with surrounding objects; tactile awareness-- ability to feel and grasp objects just tight enough to prevent slippage or crushing the objects; and 3D awareness -- ability to perceive size and depth in ways that makes object manipulation possible. Humans use these capabilities to achieve a high level of coordination needed for object manipulation. In this work, we develop techniques that equip robots with similar sensitivities towards realizing a reliable and capable home-assistant robot. In this thesis we demonstrate the importance of reasoning about the robot's workspace to enable grasping systems handle more difficult settings such as picking up moving objects while avoiding surrounding obstacles. Our method encodes the notion of reachability and uses it to generate not just stable grasps but ones that are also achievable by the robot. This reachability-aware formulation effectively expands the useable workspace of the robot enabling the robot to pick up objects from difficult-to-reach locations. While recent vision-based grasping systems work reliably well achieving pickup success rate higher than 90\% in cluttered scenes, failure cases due to calibration error, slippage and occlusion were challenging. To address this, we develop a closed-loop tactile-based improvement that uses additional tactile sensing to deal with self-occlusion (a limitation of vision-based system) and adaptively tighten the robot's grip on the object-- making the grasping system tactile-aware and more reliable. This can be used as an add-on to existing grasping systems. This adaptive tactile-based approach demonstrates the effectiveness of closed-loop feedback in the final phase of the grasping process. To achieve closed-loop manipulation all through the manipulation process, we study the value of multi-view camera systems to improve learning-based manipulation systems. Using a multi-view Q-learning formulation, we develop a learned closed-loop manipulation algorithm for precise manipulation tasks that integrates inputs from multiple static RGB cameras to overcome self-occlusion and improve 3D understanding. To conclude, we discuss some opportunities/ directions for future work.