John W. Lloyd

Alex Church, John Lloyd, Raia Hadsell et al.

Simulation has recently become key for deep reinforcement learning to safely and efficiently acquire general and complex control policies from visual and proprioceptive inputs. Tactile information is not usually considered despite its direct relation to environment interaction. In this work, we present a suite of simulated environments tailored towards tactile robotics and reinforcement learning. A simple and fast method of simulating optical tactile sensors is provided, where high-resolution contact geometry is represented as depth images. Proximal Policy Optimisation (PPO) is used to learn successful policies across all considered tasks. A data-driven approach enables translation of the current state of a real tactile sensor to corresponding simulated depth images. This policy is implemented within a real-time control loop on a physical robot to demonstrate zero-shot sim-to-real policy transfer on several physically-interactive tasks requiring a sense of touch.

Efi Psomopoulou, Nicholas Pestell, Fotios Papadopoulos et al.

This paper proposes a controller for stable grasping of unknown-shaped objects by two robotic fingers with tactile fingertips. The grasp is stabilised by rolling the fingertips on the contact surface and applying a desired grasping force to reach an equilibrium state. The validation is both in simulation and on a fully-actuated robot hand (the Shadow Modular Grasper) fitted with custom-built optical tactile sensors (based on the BRL TacTip). The controller requires the orientations of the contact surfaces, which are estimated by regressing a deep convolutional neural network over the tactile images. Overall, the grasp system is demonstrated to achieve stable equilibrium poses on various objects ranging in shape and softness, with the system being robust to perturbations and measurement errors. This approach also has promise to extend beyond grasping to stable in-hand object manipulation with multiple fingers.

John Lloyd, Nathan F. Lepora

In robots, nonprehensile manipulation operations such as pushing are a useful way of moving large, heavy or unwieldy objects, moving multiple objects at once, or reducing uncertainty in the location or pose of objects. In this study, we propose a reactive and adaptive method for robotic pushing that uses rich feedback from a high-resolution optical tactile sensor to control push movements instead of relying on analytical or data-driven models of push interactions. Specifically, we use goal-driven tactile exploration to actively search for stable pushing configurations that cause the object to maintain its pose relative to the pusher while incrementally moving the pusher and object towards the target. We evaluate our method by pushing objects across planar and curved surfaces. For planar surfaces, we show that the method is accurate and robust to variations in initial contact position/angle, object shape and start position; for curved surfaces, the performance is degraded slightly. An immediate consequence of our work is that it shows that explicit models of push interactions might be sufficient but are not necessary for this type of task. It also raises the interesting question of which aspects of the system should be modelled to achieve the best performance and generalization across a wide range of scenarios. Finally, it highlights the importance of testing on non-planar surfaces and in other more complex environments when developing new methods for robotic pushing.

Alex Church, John Lloyd, Raia Hadsell et al.

Artificial touch would seem well-suited for Reinforcement Learning (RL), since both paradigms rely on interaction with an environment. Here we propose a new environment and set of tasks to encourage development of tactile reinforcement learning: learning to type on a braille keyboard. Four tasks are proposed, progressing in difficulty from arrow to alphabet keys and from discrete to continuous actions. A simulated counterpart is also constructed by sampling tactile data from the physical environment. Using state-of-the-art deep RL algorithms, we show that all of these tasks can be successfully learnt in simulation, and 3 out of 4 tasks can be learned on the real robot. A lack of sample efficiency currently makes the continuous alphabet task impractical on the robot. To the best of our knowledge, this work presents the first demonstration of successfully training deep RL agents in the real world using observations that exclusively consist of tactile images. To aid future research utilising this environment, the code for this project has been released along with designs of the braille keycaps for 3D printing and a guide for recreating the experiments. A brief video summary is also available at https://youtu.be/eNylCA2uE_E.

Nathan F. Lepora, John Lloyd

This article illustrates the application of deep learning to robot touch by considering a basic yet fundamental capability: estimating the relative pose of part of an object in contact with a tactile sensor. We begin by surveying deep learning applied to tactile robotics, focussing on optical tactile sensors, which help bridge from deep learning for vision to touch. We then show how deep learning can be used to train accurate pose models of 3D surfaces and edges that are insensitive to nuisance variables such as motion-dependent shear. This involves including representative motions as unlabelled perturbations of the training data and using Bayesian optimization of the network and training hyperparameters to find the most accurate models. Accurate estimation of pose from touch will enable robots to safely and precisely control their physical interactions, underlying a wide range of object exploration and manipulation tasks.

Nathan F. Lepora, Alex Church, Conrad De Kerckhove et al.

Deep learning has the potential to have the impact on robot touch that it has had on robot vision. Optical tactile sensors act as a bridge between the subjects by allowing techniques from vision to be applied to touch. In this paper, we apply deep learning to an optical biomimetic tactile sensor, the TacTip, which images an array of papillae (pins) inside its sensing surface analogous to structures within human skin. Our main result is that the application of a deep CNN can give reliable edge perception and thus a robust policy for planning contact points to move around object contours. Robustness is demonstrated over several irregular and compliant objects with both tapping and continuous sliding, using a model trained only by tapping onto a disk. These results relied on using techniques to encourage generalization to tasks beyond which the model was trained. We expect this is a generic problem in practical applications of tactile sensing that deep learning will solve. A video demonstrating the approach can be found at https://www.youtube.com/watch?v=QHrGsG9AHts

Marcus Hutter, John W. Lloyd, Kee Siong Ng et al.

Automated reasoning about uncertain knowledge has many applications. One difficulty when developing such systems is the lack of a completely satisfactory integration of logic and probability. We address this problem directly. Expressive languages like higher-order logic are ideally suited for representing and reasoning about structured knowledge. Uncertain knowledge can be modeled by using graded probabilities rather than binary truth-values. The main technical problem studied in this paper is the following: Given a set of sentences, each having some probability of being true, what probability should be ascribed to other (query) sentences? A natural wish-list, among others, is that the probability distribution (i) is consistent with the knowledge base, (ii) allows for a consistent inference procedure and in particular (iii) reduces to deductive logic in the limit of probabilities being 0 and 1, (iv) allows (Bayesian) inductive reasoning and (v) learning in the limit and in particular (vi) allows confirmation of universally quantified hypotheses/sentences. We translate this wish-list into technical requirements for a prior probability and show that probabilities satisfying all our criteria exist. We also give explicit constructions and several general characterizations of probabilities that satisfy some or all of the criteria and various (counter) examples. We also derive necessary and sufficient conditions for extending beliefs about finitely many sentences to suitable probabilities over all sentences, and in particular least dogmatic or least biased ones. We conclude with a brief outlook on how the developed theory might be used and approximated in autonomous reasoning agents. Our theory is a step towards a globally consistent and empirically satisfactory unification of probability and logic.