Franziska Meier is a research scientist at the Max-Planck Institute for Intelligent Systems since August 2016. She is also a postdoctoral researcher with Dieter Fox at the University of Washington, Seattle, since February 2017. Before that she was a PhD student at the University of Southern California. She defended her thesis on “Probabilistic Machine Learning for Robotics” in 2016, under the supervision of Prof. Stefan Schaal. Prior to her PhD studies, she received her Diploma in Computer Science from the Technical University of Munich and attended the Georgia Institute of Technology as graduate student in Computer Science. Her research interests lie at the intersection of machine learning and robotics, with a special emphasis on continuous learning for intelligent agents.
Life-long Learning Machine Learning Learning Control Humanoid Robotics
One challenge towards autonomous manipulation is to cope with uncertainty and noise in the sensory-motor system of the robot and the environment. We argue that such a system requires to close feedback control loops in novel ways, using predictive models and leveraging previous experiences.
Decision making requires knowledge of some variables of interest. In the vast majority of real-world problems, these variables are latent, i.e. they cannot be observed directly and must be inferred from available measurements. To maintain an up-to-date distribution over the latent variables, past beliefs have to ...
Besides accuracy and sample efficiency, computational cost is a crucial design criterion for machine learning algorithms in real-time settings, such as control problems. An example is the modeling of robot dynamics: The sensors in a robot can produce thousands of data points per second, quickly amassing a coverage of the task relate...
In order to perform well in day to day tasks, humanoid robots need to be able to adapt to the changes in their work environments. In response to a change in the environment, the robot can pursue at least two different strategies: First, it could re-plan, a process that is often computational...
Hand-eye coordination is crucial for capable manipulation of objects. It requires to know the manipulator's and the objects' locations. These locations have to be inferred from sensory data. In this project we work with range sensors, which are wide spread in robotics and provide dense depth images.
We address the challenging problem of robotic grasping and manipulation in the presence of uncertainty. This uncertainty is due to noisy sensing, inaccurate models and hard-to-predict environment dynamics. Our approach emphasizes the importance of continuous, real-time perception and its tight integration with reactive motion genera...
IEEE Robotics and Automation Letters, 3(3):1864-1871, July 2018 (article)
We address the challenging problem of robotic grasping and manipulation in the presence of uncertainty. This uncertainty is due to noisy sensing, inaccurate models and hard-to-predict environment dynamics. Our approach emphasizes the importance of continuous, real-time perception and its tight integration with reactive motion generation methods. We present a fully integrated system where real-time object and robot tracking as well as ambient world modeling provides the necessary input to feedback controllers and continuous motion optimizers. Specifically, they provide attractive and repulsive potentials based on which the controllers and motion optimizer can online compute movement policies at different time intervals. We extensively evaluate the proposed system on a real robotic platform in four scenarios that exhibit either challenging workspace geometry or a dynamic environment. We compare the proposed integrated system with a more traditional sense-plan-act approach that is still widely used. In 333 experiments, we show the robustness and accuracy of the proposed system.
In Proceedings of the IEEE International Conference on Robotics and Automation (ICRA) 2018, IEEE, International Conference on Robotics and Automation, May 2018, accepted (inproceedings)
The promise of learning to learn for robotics rests on the hope that by extracting some information about the learning process itself we can speed up subsequent similar learning tasks. Here, we introduce a computationally efficient online meta-learning algorithm that builds and optimizes a memory model of the optimal learning rate landscape from previously observed gradient behaviors. While performing task specific optimization, this memory of learning rates predicts how to scale currently observed gradients. After applying the gradient scaling our meta-learner updates its internal memory based on the observed effect its prediction had. Our meta-learner can be combined with any gradient-based optimizer, learns on the fly and can be transferred to new optimization tasks. In our evaluations we show that our meta-learning algorithm speeds up learning of MNIST classification and a variety of learning control tasks, either in batch or online learning settings.
In Proceedings IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), IEEE, Piscataway, NJ, USA, IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), September 2017 (inproceedings)
In Proceedings of the American Control Conference, Boston, MA, USA, July 2016 (inproceedings)
Most widely-used state estimation algorithms, such as the Extended Kalman Filter and the Unscented Kalman Filter, belong to the family of Gaussian Filters (GF). Unfortunately, GFs fail if the measurement process is modelled by a fat-tailed distribution. This is a severe limitation, because thin-tailed measurement models, such as the analytically-convenient and therefore widely-used Gaussian distribution, are sensitive to outliers. In this paper, we show that mapping the measurements into a specific feature space enables any existing GF algorithm to work with fat-tailed measurement models. We find a feature function which is optimal under certain conditions. Simulation results show that the proposed method allows for robust filtering in both linear and nonlinear systems with measurements contaminated by fat-tailed noise.
Abstract: Locally weighted regression was created as a nonparametric learning method that is computationally efficient, can learn from very large amounts of data and add data incrementally. An interesting feature of locally weighted regression is that it can work with ...
In International Conference on Humanoid Robotics, pages: 512-518, IEEE, 2014, clmc (inproceedings)
Autonomous manipulation in dynamic environments is important for robots to perform everyday tasks. For this, a manipulator should be capable of interpreting the environment and planning an appropriate movement. At least, two possible approaches exist for this in literature. Usually, a planning system is used to generate a complex movement plan that satisfies all constraints. Alternatively, a simple plan could be chosen and modified with sensory feedback to accommodate additional constraints by equipping the controller with features that remain dormant most of the time, except when specific situations arise. Dynamic Movement Primitives (DMPs) form a robust and versatile starting point for such a controller that can be modified online using a non-linear term, called the coupling term. This can prove to be a fast and reactive way of obstacle avoidance in a human-like fashion. We propose a method to learn this coupling term from human demonstrations starting with simple features and making it more robust to avoid a larger range of obstacles. We test the ability of our coupling term to model different kinds of obstacle avoidance behaviours in humans and use this learnt coupling term to avoid obstacles in a reactive manner. This line of research aims at pushing the boundary of reactive control strategies to more complex scenarios, such that complex and usually computationally more expensive planning methods can be avoided as much as possible.
In Advances in Neural Information Processing Systems 27, pages: 972-980, (Editors: Z. Ghahramani, M. Welling, C. Cortes, N.D. Lawrence and K.Q. Weinberger), 28th Annual Conference on Neural Information Processing Systems (NIPS), 2014, clmc (inproceedings)
In Proceedings of the IEEE International Conference on Intelligent Robots and Systems, pages: 2244 - 2249, IROS, 2014, clmc (inproceedings)
Model-based control is essential for compliant controland force control in many modern complex robots, like humanoidor disaster robots. Due to many unknown and hard tomodel nonlinearities, analytical models of such robots are oftenonly very rough approximations. However, modern optimizationcontrollers frequently depend on reasonably accurate models,and degrade greatly in robustness and performance if modelerrors are too large. For a long time, machine learning hasbeen expected to provide automatic empirical model synthesis,yet so far, research has only generated feasibility studies butno learning algorithms that run reliably on complex robots.In this paper, we combine two promising worlds of regressiontechniques to generate a more powerful regression learningsystem. On the one hand, locally weighted regression techniquesare computationally efficient, but hard to tune due to avariety of data dependent meta-parameters. On the other hand,Bayesian regression has rather automatic and robust methods toset learning parameters, but becomes quickly computationallyinfeasible for big and high-dimensional data sets. By reducingthe complexity of Bayesian regression in the spirit of local modellearning through variational approximations, we arrive at anovel algorithm that is computationally efficient and easy toinitialize for robust learning. Evaluations on several datasetsdemonstrate very good learning performance and the potentialfor a general regression learning tool for robotics.
In Seventeenth International Conference on Artificial Intelligence and Statistics, La Palma, Canary Islands, Fifteenth International Conference on Artificial Intelligence and Statistics , April 2012 (inproceedings)
In IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS 2011), Sept. 25-30, San Francisco, CA, 2011, clmc (inproceedings)
Segmenting complex movements into a sequence of primitives remains a difficult problem with many applications in the robotics and vision communities. In this work, we show how the movement segmentation problem can be reduced to a sequential movement recognition problem. To this end, we reformulate the orig-inal Dynamic Movement Primitive (DMP) formulation as a linear dynamical sys-tem with control inputs. Based on this new formulation, we develop an Expecta-tion-Maximization algorithm to estimate the duration and goal position of a par-tially observed trajectory. With the help of this algorithm and the assumption that a library of movement primitives is present, we present a movement seg-mentation framework. We illustrate the usefulness of the new DMP formulation on the two applications of online movement recognition and movement segmen-tation.
Our goal is to understand the principles of Perception, Action and Learning in autonomous systems that successfully interact with complex environments and to use this understanding to design future systems