Dexterous robotic manipulation tasks depend on estimating the state of in-hand objects, particularly their orientation. Although cameras have been traditionally used to estimate the object’s pose, tactile sensors have recently been studied due to their robustness against occlusions. This paper explores tactile data’s temporal information for estimating the orientation of grasped objects. The data from a compliant tactile sensor were collected using different time-window sample sizes and evaluated using neural networks with long short-term memory (LSTM) layers. Our results suggest that using a window of sensor readings improved angle estimation compared to previous works. The best window size of 40 samples achieved an average of 0.0375 for the mean absolute error (MAE) in radians, 0.0030 for the mean squared error (MSE), 0.9074 for the coefficient of determination (R${}^2$), and 0.9094 for the explained variance score (EXP), with no enhancement for larger window sizes. This work illustrates the benefits of temporal information for pose estimation and analyzes the performance behavior with varying window sizes, which can be a basis for future robotic tactile research. Moreover, it can complement underactuated designs and visual pose estimation methods.

Robots need to adapt to the complexities of acting in unstructured environments. Robotic systems need more awareness of the environment to adapt to uncertainty and variability. Although cameras have been predominantly used in robotic tasks, the limitations that come with them, such as occlusion, visibility, and lack of information, have diverted some focus to tactile sensing. Extensive datasets of the physical interactions between tactile-enabled robots are required to investigate and develop methods for performing manipulation and object exploration tasks. Therefore, this motivates us to compose a dataset of signals from Bioin-Tacto modules mounted on a robotic gripper performing extraction tasks. An operator controls a robotic gripper to extract three pegs of various complexities from their corresponding holes. This dataset contains angular velocity, linear acceleration, magnetic field intensity and direction, and pressure exerted on two tactile modules embedded in the compliant structure of the sensing module. The dataset comprises 96 extraction episodes, including data collected from a reinforcement learning agent. The dataset can be used to pre-train a reinforcement machine learning model to perform peg-in-hole tasks and to study how pretraining affects a manipulator's ability to infer tactile signals and improve the success rates of the manipulator.