Composable Energy Policies for Reactive Motion Generation and Reinforcement Learning

Reactive motion generation problems are usually solved by computing actions as a sum of policies. However, these policies are independent of each other and thus, they can have conflicting behaviors when summing their contributions together. We introduce Composable Energy Policies (CEP), a novel framework for modular reactive motion generation. CEP computes the control action by optimization over the product of a set of stochastic policies. This product of policies will provide a high probability to those actions that satisfy all the components and low probability to the others. Optimizing over the product of the policies avoids the detrimental effect of conflicting behaviors between policies choosing an action that satisfies all the objectives. Besides, we show that CEP naturally adapts to the Reinforcement Learning problem allowing us to integrate, in a hierarchical fashion, any distribution as prior, from multimodal distributions to non-smooth distributions and learn a new policy given them.

Value Iteration in Continuous Actions, States and Time

Classical value iteration approaches are not applicable to environments with continuous states and actions. For such environments, the states and actions are usually discretized, which leads to an exponential increase in computational complexity. In this paper, we propose continuous fitted value iteration (cFVI). This algorithm enables dynamic programming for continuous states and actions with a known dynamics model. Leveraging the continuous-time formulation, the optimal policy can be derived for non-linear control-affine dynamics. This closed-form solution enables the efficient extension of value iteration to continuous environments. We show in non-linear control experiments that the dynamic programming solution obtains the same quantitative performance as deep reinforcement learning methods in simulation but excels when transferred to the physical system. The policy obtained by cFVI is more robust to changes in the dynamics despite using only a deterministic model and without explicitly incorporating robustness in the optimization. Videos of the physical system are available at \url{https://sites.google.com/view/value-iteration}.

Benchmarking Structured Policies and Policy Optimization for Real-World Dexterous Object Manipulation

Dexterous manipulation is a challenging and important problem in robotics. While data-driven methods are a promising approach, current benchmarks require simulation or extensive engineering support due to the sample inefficiency of popular methods. We present benchmarks for the TriFinger system, an open-source robotic platform for dexterous manipulation and the focus of the 2020 Real Robot Challenge. The benchmarked methods, which were successful in the challenge, can be generally described as structured policies, as they combine elements of classical robotics and modern policy optimization. This inclusion of inductive biases facilitates sample efficiency, interpretability, reliability and high performance. The key aspects of this benchmarking is validation of the baselines across both simulation and the real system, thorough ablation study over the core features of each solution, and a retrospective analysis of the challenge as a manipulation benchmark. The code and demo videos for this work can be found on our website (https://sites.google.com/view/benchmark-rrc).

Reinforcement Learning using Guided Observability

Due to recent breakthroughs, reinforcement learning (RL) has demonstrated impressive performance in challenging sequential decision-making problems. However, an open question is how to make RL cope with partial observability which is prevalent in many real-world problems. Contrary to contemporary RL approaches, which focus mostly on improved memory representations or strong assumptions about the type of partial observability, we propose a simple but efficient approach that can be applied together with a wide variety of RL methods. Our main insight is that smoothly transitioning from full observability to partial observability during the training process yields a high performance policy. The approach, called partially observable guided reinforcement learning (PO-GRL), allows to utilize full state information during policy optimization without compromising the optimality of the final policy. A comprehensive evaluation in discrete partially observableMarkov decision process (POMDP) benchmark problems and continuous partially observable MuJoCo and OpenAI gym tasks shows that PO-GRL improves performance. Finally, we demonstrate PO-GRL in the ball-in-the-cup task on a real Barrett WAM robot under partial observability.

Distributionally Robust Trajectory Optimization Under Uncertain Dynamics via Relative-Entropy Trust Regions

Trajectory optimization and model predictive control are essential techniques underpinning advanced robotic applications, ranging from autonomous driving to full-body humanoid control. State-of-the-art algorithms have focused on data-driven approaches that infer the system dynamics online and incorporate posterior uncertainty during planning and control. Despite their success, such approaches are still susceptible to catastrophic errors that may arise due to statistical learning biases, unmodeled disturbances or even directed adversarial attacks. In this paper, we tackle the problem of dynamics mismatch and propose a distributionally robust optimal control formulation that alternates between two relative-entropy trust region optimization problems. Our method finds the worst-case maximum-entropy Gaussian posterior over the dynamics parameters and the corresponding robust optimal policy. We show that our approach admits a closed-form backward-pass for a certain class of systems and demonstrate the resulting robustness on linear and nonlinear numerical examples.

SKID RAW: Skill Discovery from Raw Trajectories

Integrating robots in complex everyday environments requires a multitude of problems to be solved. One crucial feature among those is to equip robots with a mechanism for teaching them a new task in an easy and natural way. When teaching tasks that involve sequences of different skills, with varying order and number of these skills, it is desirable to only demonstrate full task executions instead of all individual skills. For this purpose, we propose a novel approach that simultaneously learns to segment trajectories into reoccurring patterns and the skills to reconstruct these patterns from unlabelled demonstrations without further supervision. Moreover, the approach learns a skill conditioning that can be used to understand possible sequences of skills, a practical mechanism to be used in, for example, human-robot-interactions for a more intelligent and adaptive robot behaviour. The Bayesian and variational inference based approach is evaluated on synthetic and real human demonstrations with varying complexities and dimensionality, showing the successful learning of segmentations and skill libraries from unlabelled data.

Model Predictive Actor-Critic: Accelerating Robot Skill Acquisition with Deep Reinforcement Learning

Substantial advancements to model-based reinforcement learning algorithms have been impeded by the model-bias induced by the collected data, which generally hurts performance. Meanwhile, their inherent sample efficiency warrants utility for most robot applications, limiting potential damage to the robot and its environment during training. Inspired by information theoretic model predictive control and advances in deep reinforcement learning, we introduce Model Predictive Actor-Critic (MoPAC), a hybrid model-based/model-free method that combines model predictive rollouts with policy optimization as to mitigate model bias. MoPAC leverages optimal trajectories to guide policy learning, but explores via its model-free method, allowing the algorithm to learn more expressive dynamics models. This combination guarantees optimal skill learning up to an approximation error and reduces necessary physical interaction with the environment, making it suitable for real-robot training. We provide extensive results showcasing how our proposed method generally outperforms current state-of-the-art and conclude by evaluating MoPAC for learning on a physical robotic hand performing valve rotation and finger gaiting--a task that requires grasping, manipulation, and then regrasping of an object.

Advancing Trajectory Optimization with Approximate Inference: Exploration, Covariance Control and Adaptive Risk

Discrete-time stochastic optimal control remains a challenging problem for general, nonlinear systems under significant uncertainty, with practical solvers typically relying on the certainty equivalence assumption, replanning and/or extensive regularization. Control as inference is an approach that frames stochastic control as an equivalent inference problem, and has demonstrated desirable qualities over existing methods, namely in exploration and regularization. We look specifically at the input inference for control (i2c) algorithm, and derive three key characteristics that enable advanced trajectory optimization: An `expert' linear Gaussian controller that combines the benefits of open-loop optima and closed-loop variance reduction when optimizing for nonlinear systems, inherent adaptive risk sensitivity from the inference formulation, and covariance control functionality with only a minor algorithmic adjustment.

Extended Task and Motion Planning of Long-horizon Robot Manipulation

Task and Motion Planning (TAMP) requires the integration of symbolic reasoning with metric motion planning that accounts for the robot's actions' geometric feasibility. This hierarchical structure inevitably prevents the symbolic planners from accessing the environment's low-level geometric description, vital to the problem's solution. Most TAMP approaches fail to provide feasible solutions when there is missing knowledge about the environment at the symbolic level. The incapability of devising alternative high-level plans leads existing planners to a dead end. We propose a novel approach for decision-making on extended decision spaces over plan skeletons and action parameters. We integrate top-k planning for constructing an explicit skeleton space, where a skeleton planner generates a variety of candidate skeleton plans. Moreover, we effectively combine this skeleton space with the resultant motion parameter spaces into a single extended decision space. Accordingly, we use Monte-Carlo Tree Search (MCTS) to ensure an exploration-exploitation balance at each decision node and optimize globally to produce minimum-cost solutions. The proposed seamless combination of symbolic top-k planning with streams, with the proved optimality of MCTS, leads to a powerful planning algorithm that can handle the combinatorial complexity of long-horizon manipulation tasks. We empirically evaluate our proposed algorithm in challenging manipulation tasks with different domains that require multi-stage decisions and show how our method can overcome dead-ends through its effective alternate plans compared to its most competitive baseline method.

Learning Human-like Hand Reaching for Human-Robot Handshaking

One of the first and foremost non-verbal interactions that humans perform is a handshake. It has an impact on first impressions as touch can convey complex emotions. This makes handshaking an important skill for the repertoire of a social robot. In this paper, we present a novel framework for learning human-robot handshaking behaviours for humanoid robots solely using third-person human-human interaction data. This is especially useful for non-backdrivable robots that cannot be taught by demonstrations via kinesthetic teaching. Our approach can be easily executed on different humanoid robots. This removes the need for re-training, which is especially tedious when training with human-interaction partners. We show this by applying the learnt behaviours on two different humanoid robots with similar degrees of freedom but different shapes and control limits.