A probabilistic graphical model is proposed, modeling the joint model parameter and multiplier evolution, with a hypervolume based likelihood, promoting multi-objective descent in structural risk minimization. We address multi-objective model parameter optimization via a surrogate single objective penalty loss with time-varying multipliers, equivalent to online scheduling of loss landscape. The multi-objective descent goal is dispatched hierarchically into a series of constraint optimization sub-problems with shrinking bounds according to Pareto dominance. The bound serves as setpoint for the low-level multiplier controller to schedule loss landscapes via output feedback of each loss term. Our method forms closed loop of model parameter dynamic, circumvents excessive memory requirements and extra computational burden of existing multi-objective deep learning methods, and is robust against controller hyperparameter variation, demonstrated on domain generalization tasks.

Most recent successes in robot reinforcement learning involve learning a specialized single-task agent. However, robots capable of performing multiple tasks can be much more valuable in real-world applications. Multi-task reinforcement learning can be very challenging due to the increased sample complexity and the potentially conflicting task objectives. Previous work on this topic is dominated by model-free approaches. The latter can be very sample inefficient even when learning specialized single-task agents. In this work, we focus on model-based multi-task reinforcement learning. We propose a method for learning multi-task visual world models, leveraging pre-trained language models to extract semantically meaningful task representations. These representations are used by the world model and policy to reason about task similarity in dynamics and behavior. Our results highlight the benefits of using language-driven task representations for world models and a clear advantage of model-based multi-task learning over the more common model-free paradigm.

Measuring the similarity between data points often requires domain knowledge, which can in parts be compensated by relying on unsupervised methods such as latent-variable models, where similarity/distance is estimated in a more compact latent space. Prevalent is the use of the Euclidean metric, which has the drawback of ignoring information about similarity of data stored in the decoder, as captured by the framework of Riemannian geometry. We propose an extension to the framework of variational auto-encoders allows learning flat latent manifolds, where the Euclidean metric is a proxy for the similarity between data points. This is achieved by defining the latent space as a Riemannian manifold and by regularising the metric tensor to be a scaled identity matrix. Additionally, we replace the compact prior typically used in variational auto-encoders with a recently presented, more expressive hierarchical one—and formulate the learning problem as a constrained optimisation problem. We evaluate our method on a range of data-sets, including a video-tracking benchmark, where the performance of our unsupervised approach nears that of state-of-the-art supervised approaches, while retaining the computational efficiency of straight-line-based approaches.

We propose to learn a hierarchical prior in the context of variational autoencoders to avoid the over-regularisation resulting from a standard normal prior distribution. To incentivise an informative latent representation of the data, we formulate the learning problem as a constrained optimisation problem by extending the Taming VAEs framework to two-level hierarchical models. We introduce a graph-based interpolation method, which shows that the topology of the learned latent representation corresponds to the topology of the data manifold---and present several examples, where desired properties of latent representation such as smoothness and simple explanatory factors are learned by the prior.

Neural samplers such as variational autoencoders (VAEs) or generative adversarial networks (GANs) approximate distributions by transforming samples from a simple random source—the latent space—to samples from a more complex distribution represented by a dataset. While the manifold hypothesis implies that a dataset contains large regions of low density, the training criterions of VAEs and GANs will make the latent space densely covered. Consequently points that are separated by low-density regions in observation space will be pushed together in latent space, making stationary distances poor proxies for similarity. We transfer ideas from Riemannian geometry to this setting, letting the distance between two points be the shortest path on a Riemannian manifold induced by the transformation. The method yields a principled distance measure, provides a tool for visual inspection of deep generative models, and an alternative to linear interpolation in latent space. In addition, it can be applied for robot movement generalization using previously learned skills. The method is evaluated on a synthetic dataset with known ground truth; on a simulated robot arm dataset; on human motion capture data; and on a generative model of handwritten digits.

Robots can rapidly acquire new skills from demonstrations. However, during generalisation of skills or transitioning across fundamentally different skills, it is unclear whether the robot has the necessary knowledge to perform the task. Failing to detect missing information often leads to abrupt movements or to collisions with the environment. Active learning can quantify the uncertainty of performing the task and, in general, locate regions of missing information. We introduce a novel algorithm for active learning and demonstrate its utility for generating smooth trajectories. Our approach is based on deep generative models and metric learning in latent spaces. It relies on the Jacobian of the likelihood to detect non-smooth transitions in the latent space, i.e., transitions that lead to abrupt changes in the movement of the robot. When non-smooth transitions are detected, our algorithm asks for an additional demonstration from that specific region. The newly acquired knowledge modifies the data manifold and allows for learning a latent representation for generating smooth movements. We demonstrate the efficacy of our approach on generalising elementary skills, transitioning across different skills, and implicitly avoiding collisions with the environment. For our experiments, we use a simulated pendulum where we observe its motion from images and a 7-DoF anthropomorphic arm.

For many tasks, tactile or visual feedback is helpful or even crucial. However, designing controllers that take such high-dimensional feedback into account is non-trivial. Therefore, robots should be able to learn tactile skills through trial and error by using reinforcement learning algorithms. The input domain for such tasks, however, might include strongly correlated or non-relevant dimensions, making it hard to specify a suitable metric on such domains. Auto-encoders specialize in finding compact representations, where defining such a metric is likely to be easier. Therefore, we propose a reinforcement learning algorithm that can learn non-linear policies in continuous state spaces, which leverages representations learned using auto-encoders. We first evaluate this method on a simulated toy-task with visual input. Then, we validate our approach on a real-robot tactile stabilization task.

Robust fingertip force detection from fingernail image is a critical strategy that can be applied in many areas. However, prior research fixed many variables that influence the finger color change. This paper analyzes the effect of the finger joint on the force detection in order to deal with the constrained finger position setting. A force estimator method is designed: a model to predict the fingertip force from finger joints measured from 2D cameras and 3 rectangular markers in cooperation with the fingernail images are trained. Then the error caused by the color changes of the joint bending can be avoided. This strategy is a significant step forward from a finger force estimator that requires tedious finger joint setting. The approach is evaluated experimentally. The result shows that it increases the accuracy over 10% for the force in conditions of the finger joint free movement. The estimator is used to demonstrate lifting and replacing objects with various weights.

Estimating human fingertip forces is required to understand force distribution in grasping and manipulation. Human grasping behavior can then be used to develop force-and impedance-based grasping and manipulation strategies for robotic hands. However, estimating human grip force naturally is only possible with instrumented objects or unnatural gloves, thus greatly limiting the type of objects used. In this paper we describe an approach which uses images of the human fingertip to reconstruct grip force and torque at the finger. Our approach does not use finger-mounted equipment, but instead a steady camera observing the fingers of the hand from a distance. This allows for finger force estimation without any physical interference with the hand or object itself, and is therefore universally applicable. We construct a 3-dimensional finger model from 2D images. Convolutional Neural Networks (CNN) are used to predict the 2D image to a 3D model transformation matrix. Two methods of CNN are designed for separate and combined outputs of orientation and position. After learning, our system shows an alignment accuracy over 98% on unknown data. In the final step, a Gaussian process estimates finger force and torque from the aligned images based on color changes and deformations of the nail and its surrounding skin. Experimental results shows that the accuracy achieves about 95% in the force estimation and 90% in the torque.

Teleoperated grasping requires the abilities to follow the intended trajectory from the user and autonomously search for a suitable pre-grasp pose relative to the object of interest. Challenges include dealing with uncertainty due to the noise of teleoperator, human elements and calibration errors in the sensors. To address these challenges, an effective and robust algorithm is introduced to assist grasping during teleoperation. Although without premature object contact or regrasping strategies, the algorithm enable the robot to perform online adjustments to reach a pre-grasp pose for a final grasping. We use three infra-red (IR) sensors that are mounted on the robot hand, and design an algorithm that controls the robot hand to grasp objects using the information from the sensors readings and the interface component. Finally, a series of experiments demonstrate that the system is robust when grasping a wide range of objects and even tracks mobile objects. Empirical data from a 5-subject user study allows us to tune the relative contributions from the IR sensors and the interface component, so as to achieve a balance of grasp assistance and teleoperation.

In order to provide a user-friendly system with simple operation command to grasp different objects successfully, this paper describes a combined approach of real time remote vision-based teleoperation and autonomy for a human-aided robotic grasping. In the teleoperation process, motion tracking is carried out by Kinect in real time to detect the positions of the human shoulder, elbow and hand joints such that the robot can imitate the human. Hand gestures are recognized and used to activate autonomous grasping, which can save time and generate more natural grasping poses. In our system, the robot fulfills some special tasks such as picking up objects using easy commands with Kinect as object sensor. Experiment results show that it is effective and user-friendly.