Purpose of Review This review provides a comprehensive overview of machine learning approaches for vision-based robotic grasping and manipulation. Current trends and developments as well as various criteria for categorization of approaches are provided. Recent Findings Model-free approaches are attractive due to their generalization capabilities to novel objects, but are mostly limited to top-down grasps and do not allow a precise object placement which can limit their applicability. In contrast, model-based methods allow a precise placement and aim for an automatic configuration without any human intervention to enable a fast and easy deployment. Summary Both approaches to robotic grasping and manipulation with and without object-specific knowledge are discussed. Due to the large amount of data required to train AI-based approaches, simulations are an attractive choice for robot learning. This article also gives an overview of techniques and achievements in transfers from simulations to the real world.

No abstract

In a cable-driven parallel robot, elastic cables are used to manipulate the end effector in the workspace. In this paper we present a dynamic analysis and system identification for the complete actuator unit of a cable robot including servo controller, winch, cable, cable force sensor and field bus communication. We establish a second-order system with dead time as an analagous model. Based on this investigation, we propose the design and stability analysis of a cable force controller. We present the implementation of feed-forward and integral controllers based on a stiffness model of the cables. As the platform position is not observable the challenge is to control the cable force while maintaining the positional accuracy. Experimental evaluation of the force controller shows, that the absolute positional accuracy is even improved

In a cable-driven parallel robot, elastic cables are used to manipulate the end effector within the workspace. Cable force measurement is necessary for several control algorithms like cable force control, contact control, or load identification. The cable force sensor can be placed directly at the connection point on the platform or somewhere along the cable using pulleys. The pulleys between the force sensor and the platform disturb the force measurement accuracy due to friction. This paper deals with modeling and compensation of the friction. The friction behavior in the drive train with focus on the effects of the pulleys is non-trivial, as the cable movement consists of microscopic and macroscopic movements and standstills. Friction models from Coulomb and Dahl are adapted to deal with the pulley friction. The experimental evaluation showed an improvement of 70% with respect to the uncompensated case

No abstract