No abstract

Robots operating in unstructured environments must be capable of safely handling unexpected collisions with objects existing in the surrounding area, possibly without stopping the task execution. This paper proposes a tactile feedback control law allowing the robot to apply bounded contact forces in reaction to physical collisions while performing a task. As a use-case scenario, the problem of driving a robot arm through obstacles to reach a known target position in the space is considered. Intensities and locations of multiple and simultaneous contacts between the robot and the environment are detected using large area tactile sensors covering the robot body. It will be shown that the robot is capable of controlling the end-effector position and of reacting to unexpected collisions by regulating the interaction forces applied to the environment. The method has been validated on a Baxter robot partially covered with 3495 distributed tactile elements, operating in an unknown environment.

No abstract

In this paper we present a performance analysis of a perceptual architecture for industrial robots based on tactile sensing. We have developed a human-robot interface based on sensing devices which can be fixed on the robot body or the gripper. The devices are cylindershaped handles covered with tactile sensors, which can be intuitively grasped by an operator. Tactile data are fed into a neural network in order to recognize human touch during grasp, thus providing an enabling command for the control system. We provide an inference time comparison between two computing architectures: a desktop workstation with an high-performance GPU and an embedded solution based on the NVIDIA Jetson Nano board. We also compare inference time obtained from three instances of the neural network, compiled with three different engines: Keras, TensorRT floating-point 16 and TensorRT floating-point 32, showing that numerically optimized models allow to perform inference also on the embedded board without violating timing constraints imposed by data acquisition. Moreover, we assess the robustness of touch recognition when the user is wearing work gloves, showing that the difference from the bare-hand case is negligible.

No abstract