Header logo is


2019


Thumb xl paper images.005
Learning to Explore in Motion and Interaction Tasks

Bogdanovic, M., Righetti, L.

IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), IEEE, November 2019 (conference)

Abstract
Model free reinforcement learning suffers from the high sampling complexity inherent to robotic manipulation or locomotion tasks. Most successful approaches typically use random sampling strategies which leads to slow policy convergence. In this paper we present a novel approach for efficient exploration that leverages previously learned tasks. We exploit the fact that the same system is used across many tasks and build a generative model for exploration based on data from previously solved tasks to improve learning new tasks. The approach also enables continuous learning of improved exploration strategies as novel tasks are learned. Extensive simulations on a robot manipulator performing a variety of motion and contact interaction tasks demonstrate the capabilities of the approach. In particular, our experiments suggest that the exploration strategy can more than double learning speed, especially when rewards are sparse. Moreover, the algorithm is robust to task variations and parameter tuning, making it beneficial for complex robotic problems.

mg

arXiv [BibTex]

2019


arXiv [BibTex]


Thumb xl paper images.006
Learning Variable Impedance Control for Contact Sensitive Tasks

Bogdanovic, M., Khadiv, M., Righetti, L.

arXiv preprint, arXiv:1907.07500, July 2019 (article)

Abstract
Reinforcement learning algorithms have shown great success in solving different problems ranging from playing video games to robotics. However, they struggle to solve delicate robotic problems, especially those involving contact interactions. Though in principle a policy outputting joint torques should be able to learn these tasks, in practice we see that they have difficulty to robustly solve the problem without any structure in the action space. In this paper, we investigate how the choice of action space can give robust performance in presence of contact uncertainties. We propose to learn a policy that outputs impedance and desired position in joint space as a function of system states without imposing any other structure to the problem. We compare the performance of this approach to torque and position control policies under different contact uncertainties. Extensive simulation results on two different systems, a hopper (floating-base) with intermittent contacts and a manipulator (fixed-base) wiping a table, show that our proposed approach outperforms policies outputting torque or position in terms of both learning rate and robustness to environment uncertainty.

mg

[BibTex]


no image
Variational Autoencoders Recover PCA Directions (by Accident)

Rolinek, M., Zietlow, D., Martius, G.

In Proceedings IEEE Conf. on Computer Vision and Pattern Recognition (CVPR), IEEE International Conference on Computer Vision and Pattern Recognition (CVPR) 2019, June 2019 (inproceedings)

Abstract
The Variational Autoencoder (VAE) is a powerful architecture capable of representation learning and generative modeling. When it comes to learning interpretable (disentangled) representations, VAE and its variants show unparalleled performance. However, the reasons for this are unclear, since a very particular alignment of the latent embedding is needed but the design of the VAE does not encourage it in any explicit way. We address this matter and offer the following explanation: the diagonal approximation in the encoder together with the inherent stochasticity force local orthogonality of the decoder. The local behavior of promoting both reconstruction and orthogonality matches closely how the PCA embedding is chosen. Alongside providing an intuitive understanding, we justify the statement with full theoretical analysis as well as with experiments.

al

arXiv link (url) Project Page [BibTex]

arXiv link (url) Project Page [BibTex]


Thumb xl paper images.007
Leveraging Contact Forces for Learning to Grasp

Merzic, H., Bogdanovic, M., Kappler, D., Righetti, L., Bohg, J.

In Proceedings of the IEEE International Conference on Robotics and Automation (ICRA) 2019, IEEE, International Conference on Robotics and Automation, May 2019 (inproceedings)

Abstract
Grasping objects under uncertainty remains an open problem in robotics research. This uncertainty is often due to noisy or partial observations of the object pose or shape. To enable a robot to react appropriately to unforeseen effects, it is crucial that it continuously takes sensor feedback into account. While visual feedback is important for inferring a grasp pose and reaching for an object, contact feedback offers valuable information during manipulation and grasp acquisition. In this paper, we use model-free deep reinforcement learning to synthesize control policies that exploit contact sensing to generate robust grasping under uncertainty. We demonstrate our approach on a multi-fingered hand that exhibits more complex finger coordination than the commonly used two- fingered grippers. We conduct extensive experiments in order to assess the performance of the learned policies, with and without contact sensing. While it is possible to learn grasping policies without contact sensing, our results suggest that contact feedback allows for a significant improvement of grasping robustness under object pose uncertainty and for objects with a complex shape.

am mg

video arXiv [BibTex]

video arXiv [BibTex]


no image
Control What You Can: Intrinsically Motivated Task-Planning Agent

Blaes, S., Vlastelica, M., Zhu, J., Martius, G.

In Advances in Neural Information Processing (NeurIPS’19), pages: 12520-12531, Curran Associates, Inc., NeurIPS'19, 2019 (inproceedings)

Abstract
We present a novel intrinsically motivated agent that learns how to control the environment in the fastest possible manner by optimizing learning progress. It learns what can be controlled, how to allocate time and attention, and the relations between objects using surprise based motivation. The effectiveness of our method is demonstrated in a synthetic as well as a robotic manipulation environment yielding considerably improved performance and smaller sample complexity. In a nutshell, our work combines several task-level planning agent structures (backtracking search on task graph, probabilistic road-maps, allocation of search efforts) with intrinsic motivation to achieve learning from scratch.

al

link (url) Project Page [BibTex]

link (url) Project Page [BibTex]


Thumb xl systemillustration
Autonomous Identification and Goal-Directed Invocation of Event-Predictive Behavioral Primitives

Gumbsch, C., Butz, M. V., Martius, G.

IEEE Transactions on Cognitive and Developmental Systems, 2019 (article)

Abstract
Voluntary behavior of humans appears to be composed of small, elementary building blocks or behavioral primitives. While this modular organization seems crucial for the learning of complex motor skills and the flexible adaption of behavior to new circumstances, the problem of learning meaningful, compositional abstractions from sensorimotor experiences remains an open challenge. Here, we introduce a computational learning architecture, termed surprise-based behavioral modularization into event-predictive structures (SUBMODES), that explores behavior and identifies the underlying behavioral units completely from scratch. The SUBMODES architecture bootstraps sensorimotor exploration using a self-organizing neural controller. While exploring the behavioral capabilities of its own body, the system learns modular structures that predict the sensorimotor dynamics and generate the associated behavior. In line with recent theories of event perception, the system uses unexpected prediction error signals, i.e., surprise, to detect transitions between successive behavioral primitives. We show that, when applied to two robotic systems with completely different body kinematics, the system manages to learn a variety of complex behavioral primitives. Moreover, after initial self-exploration the system can use its learned predictive models progressively more effectively for invoking model predictive planning and goal-directed control in different tasks and environments.

al

arXiv PDF video link (url) DOI Project Page [BibTex]


no image
Machine Learning for Haptics: Inferring Multi-Contact Stimulation From Sparse Sensor Configuration

Sun, H., Martius, G.

Frontiers in Neurorobotics, 13, pages: 51, 2019 (article)

Abstract
Robust haptic sensation systems are essential for obtaining dexterous robots. Currently, we have solutions for small surface areas such as fingers, but affordable and robust techniques for covering large areas of an arbitrary 3D surface are still missing. Here, we introduce a general machine learning framework to infer multi-contact haptic forces on a 3D robot’s limb surface from internal deformation measured by only a few physical sensors. The general idea of this framework is to predict first the whole surface deformation pattern from the sparsely placed sensors and then to infer number, locations and force magnitudes of unknown contact points. We show how this can be done even if training data can only be obtained for single-contact points using transfer learning at the example of a modified limb of the Poppy robot. With only 10 strain-gauge sensors we obtain a high accuracy also for multiple-contact points. The method can be applied to arbitrarily shaped surfaces and physical sensor types, as long as training data can be obtained.

al

link (url) DOI [BibTex]


no image
Robust Humanoid Locomotion Using Trajectory Optimization and Sample-Efficient Learning

Yeganegi, M. H., Khadiv, M., Moosavian, S. A. A., Zhu, J., Prete, A. D., Righetti, L.

Proceedings International Conference on Humanoid Robots, IEEE, 2019 IEEE-RAS International Conference on Humanoid Robots, 2019 (conference)

Abstract
Trajectory optimization (TO) is one of the most powerful tools for generating feasible motions for humanoid robots. However, including uncertainties and stochasticity in the TO problem to generate robust motions can easily lead to intractable problems. Furthermore, since the models used in TO have always some level of abstraction, it can be hard to find a realistic set of uncertainties in the model space. In this paper we leverage a sample-efficient learning technique (Bayesian optimization) to robustify TO for humanoid locomotion. The main idea is to use data from full-body simulations to make the TO stage robust by tuning the cost weights. To this end, we split the TO problem into two phases. The first phase solves a convex optimization problem for generating center of mass (CoM) trajectories based on simplified linear dynamics. The second stage employs iterative Linear-Quadratic Gaussian (iLQG) as a whole-body controller to generate full body control inputs. Then we use Bayesian optimization to find the cost weights to use in the first stage that yields robust performance in the simulation/experiment, in the presence of different disturbance/uncertainties. The results show that the proposed approach is able to generate robust motions for different sets of disturbances and uncertainties.

mg

https://arxiv.org/abs/1907.04616 [BibTex]

https://arxiv.org/abs/1907.04616 [BibTex]

2013


no image
AGILITY – Dynamic Full Body Locomotion and Manipulation with Autonomous Legged Robots

Hutter, M., Bloesch, M., Buchli, J., Semini, C., Bazeille, S., Righetti, L., Bohg, J.

In 2013 IEEE International Symposium on Safety, Security, and Rescue Robotics (SSRR), pages: 1-4, IEEE, Linköping, Sweden, 2013 (inproceedings)

mg

link (url) DOI [BibTex]

2013


link (url) DOI [BibTex]


no image
Learning Objective Functions for Manipulation

Kalakrishnan, M., Pastor, P., Righetti, L., Schaal, S.

In 2013 IEEE International Conference on Robotics and Automation, IEEE, Karlsruhe, Germany, 2013 (inproceedings)

Abstract
We present an approach to learning objective functions for robotic manipulation based on inverse reinforcement learning. Our path integral inverse reinforcement learning algorithm can deal with high-dimensional continuous state-action spaces, and only requires local optimality of demonstrated trajectories. We use L 1 regularization in order to achieve feature selection, and propose an efficient algorithm to minimize the resulting convex objective function. We demonstrate our approach by applying it to two core problems in robotic manipulation. First, we learn a cost function for redundancy resolution in inverse kinematics. Second, we use our method to learn a cost function over trajectories, which is then used in optimization-based motion planning for grasping and manipulation tasks. Experimental results show that our method outperforms previous algorithms in high-dimensional settings.

am mg

link (url) DOI [BibTex]

link (url) DOI [BibTex]


no image
Information Driven Self-Organization of Complex Robotic Behaviors

Martius, G., Der, R., Ay, N.

PLoS ONE, 8(5):e63400, Public Library of Science, 2013 (article)

al

link (url) DOI [BibTex]

link (url) DOI [BibTex]


no image
Optimal distribution of contact forces with inverse-dynamics control

Righetti, L., Buchli, J., Mistry, M., Kalakrishnan, M., Schaal, S.

The International Journal of Robotics Research, 32(3):280-298, March 2013 (article)

Abstract
The development of legged robots for complex environments requires controllers that guarantee both high tracking performance and compliance with the environment. More specifically the control of the contact interaction with the environment is of crucial importance to ensure stable, robust and safe motions. In this contribution we develop an inverse-dynamics controller for floating-base robots under contact constraints that can minimize any combination of linear and quadratic costs in the contact constraints and the commands. Our main result is the exact analytical derivation of the controller. Such a result is particularly relevant for legged robots as it allows us to use torque redundancy to directly optimize contact interactions. For example, given a desired locomotion behavior, we can guarantee the minimization of contact forces to reduce slipping on difficult terrains while ensuring high tracking performance of the desired motion. The main advantages of the controller are its simplicity, computational efficiency and robustness to model inaccuracies. We present detailed experimental results on simulated humanoid and quadruped robots as well as a real quadruped robot. The experiments demonstrate that the controller can greatly improve the robustness of locomotion of the robots.1

am mg

link (url) DOI [BibTex]

link (url) DOI [BibTex]


no image
Linear combination of one-step predictive information with an external reward in an episodic policy gradient setting: a critical analysis

Zahedi, K., Martius, G., Ay, N.

Frontiers in Psychology, 4(801), 2013 (article)

Abstract
One of the main challenges in the field of embodied artificial intelligence is the open-ended autonomous learning of complex behaviours. Our approach is to use task-independent, information-driven intrinsic motivation(s) to support task-dependent learning. The work presented here is a preliminary step in which we investigate the predictive information (the mutual information of the past and future of the sensor stream) as an intrinsic drive, ideally supporting any kind of task acquisition. Previous experiments have shown that the predictive information (PI) is a good candidate to support autonomous, open-ended learning of complex behaviours, because a maximisation of the PI corresponds to an exploration of morphology- and environment-dependent behavioural regularities. The idea is that these regularities can then be exploited in order to solve any given task. Three different experiments are presented and their results lead to the conclusion that the linear combination of the one-step PI with an external reward function is not generally recommended in an episodic policy gradient setting. Only for hard tasks a great speed-up can be achieved at the cost of an asymptotic performance lost.

al

link (url) DOI [BibTex]


no image
Robustness of guided self-organization against sensorimotor disruptions

Martius, G.

Advances in Complex Systems, 16(02n03):1350001, 2013 (article)

Abstract
Self-organizing processes are crucial for the development of living beings. Practical applications in robots may benefit from the self-organization of behavior, e.g.~to increase fault tolerance and enhance flexibility, provided that external goals can also be achieved. We present results on the guidance of self-organizing control by visual target stimuli and show a remarkable robustness to sensorimotor disruptions. In a proof of concept study an autonomous wheeled robot is learning an object finding and ball-pushing task from scratch within a few minutes in continuous domains. The robustness is demonstrated by the rapid recovery of the performance after severe changes of the sensor configuration.

al

DOI [BibTex]

DOI [BibTex]


no image
Controlled Reduction with Unactuated Cyclic Variables: Application to 3D Bipedal Walking with Passive Yaw Rotation

Gregg, R., Righetti, L.

IEEE Transactions on Automatic Control, 58(10):2679-2685, October 2013 (article)

Abstract
This technical note shows that viscous damping can shape momentum conservation laws in a manner that stabilizes yaw rotation and enables steering for underactuated 3D walking. We first show that unactuated cyclic variables can be controlled by passively shaped conservation laws given a stabilizing controller in the actuated coordinates. We then exploit this result to realize controlled geometric reduction with multiple unactuated cyclic variables. We apply this underactuated control strategy to a five-link 3D biped to produce exponentially stable straight-ahead walking and steering in the presence of passive yawing.

mg

link (url) DOI [BibTex]

link (url) DOI [BibTex]


no image
Learning Task Error Models for Manipulation

Pastor, P., Kalakrishnan, M., Binney, J., Kelly, J., Righetti, L., Sukhatme, G. S., Schaal, S.

In 2013 IEEE Conference on Robotics and Automation, IEEE, Karlsruhe, Germany, 2013 (inproceedings)

Abstract
Precise kinematic forward models are important for robots to successfully perform dexterous grasping and manipulation tasks, especially when visual servoing is rendered infeasible due to occlusions. A lot of research has been conducted to estimate geometric and non-geometric parameters of kinematic chains to minimize reconstruction errors. However, kinematic chains can include non-linearities, e.g. due to cable stretch and motor-side encoders, that result in significantly different errors for different parts of the state space. Previous work either does not consider such non-linearities or proposes to estimate non-geometric parameters of carefully engineered models that are robot specific. We propose a data-driven approach that learns task error models that account for such unmodeled non-linearities. We argue that in the context of grasping and manipulation, it is sufficient to achieve high accuracy in the task relevant state space. We identify this relevant state space using previously executed joint configurations and learn error corrections for those. Therefore, our system is developed to generate subsequent executions that are similar to previous ones. The experiments show that our method successfully captures the non-linearities in the head kinematic chain (due to a counterbalancing spring) and the arm kinematic chains (due to cable stretch) of the considered experimental platform, see Fig. 1. The feasibility of the presented error learning approach has also been evaluated in independent DARPA ARM-S testing contributing to successfully complete 67 out of 72 grasping and manipulation tasks.

am mg

link (url) DOI [BibTex]

link (url) DOI [BibTex]

2009


no image
Modelling the interplay of central pattern generation and sensory feedback in the neuromuscular control of running

Daley, M., Righetti, L., Ijspeert, A.

In Comparative Biochemistry and Physiology - Part A: Molecular & Integrative Physiology. Annual Main Meeting for the Society for Experimental Biology, 153, Glasgow, Scotland, 2009 (inproceedings)

mg

link (url) DOI [BibTex]

2009


link (url) DOI [BibTex]


no image
Adaptive Frequency Oscillators and Applications

Righetti, L., Buchli, J., Ijspeert, A.

The Open Cybernetics \& Systemics Journal, 3, pages: 64-69, 2009 (article)

Abstract
In this contribution we present a generic mechanism to transform an oscillator into an adaptive frequency oscillator, which can then dynamically adapt its parameters to learn the frequency of any periodic driving signal. Adaptation is done in a dynamic way: it is part of the dynamical system and not an offline process. This mechanism goes beyond entrainment since it works for any initial frequencies and the learned frequency stays encoded in the system even if the driving signal disappears. Interestingly, this mechanism can easily be applied to a large class of oscillators from harmonic oscillators to relaxation types and strange attractors. Several practical applications of this mechanism are then presented, ranging from adaptive control of compliant robots to frequency analysis of signals and construction of limit cycles of arbitrary shape.

mg

link (url) [BibTex]

link (url) [BibTex]


no image
A Sensor-Based Learning Algorithm for the Self-Organization of Robot Behavior

Hesse, F., Martius, G., Der, R., Herrmann, J. M.

Algorithms, 2(1):398-409, 2009 (article)

Abstract
Ideally, sensory information forms the only source of information to a robot. We consider an algorithm for the self-organization of a controller. At short timescales the controller is merely reactive but the parameter dynamics and the acquisition of knowledge by an internal model lead to seemingly purposeful behavior on longer timescales. As a paradigmatic example, we study the simulation of an underactuated snake-like robot. By interacting with the real physical system formed by the robotic hardware and the environment, the controller achieves a sensitive and body-specific actuation of the robot.

al

link (url) [BibTex]

link (url) [BibTex]