Real World Offline Reinforcement Learning with Realistic Data Source

Offline reinforcement learning (ORL) holds great promise for robot learning due to its ability to learn from arbitrary pre-generated experience. However, current ORL benchmarks are almost entirely in simulation and utilize contrived datasets like replay buffers of online RL agents or sub-optimal trajectories, and thus hold limited relevance for real-world robotics. In this work (Real-ORL), we posit that data collected from safe operations of closely related tasks are more practical data sources for real-world robot learning. Under these settings, we perform an extensive (6500+ trajectories collected over 800+ robot hours and 270+ human labor hour) empirical study evaluating generalization and transfer capabilities of representative ORL methods on four real-world tabletop manipulation tasks. Our study finds that ORL and imitation learning prefer different action spaces, and that ORL algorithms can generalize from leveraging offline heterogeneous data sources and outperform imitation learning. We release our dataset and implementations at URL: https://sites.google.com/view/real-orl

Git Re-Basin: Merging Models modulo Permutation Symmetries

The success of deep learning is thanks to our ability to solve certain massive non-convex optimization problems with relative ease. Despite non-convex optimization being NP-hard, simple algorithms -- often variants of stochastic gradient descent -- exhibit surprising effectiveness in fitting large neural networks in practice. We argue that neural network loss landscapes contain (nearly) a single basin, after accounting for all possible permutation symmetries of hidden units. We introduce three algorithms to permute the units of one model to bring them into alignment with units of a reference model. This transformation produces a functionally equivalent set of weights that lie in an approximately convex basin near the reference model. Experimentally, we demonstrate the single basin phenomenon across a variety of model architectures and datasets, including the first (to our knowledge) demonstration of zero-barrier linear mode connectivity between independently trained ResNet models on CIFAR-10 and CIFAR-100. Additionally, we identify intriguing phenomena relating model width and training time to mode connectivity across a variety of models and datasets. Finally, we discuss shortcomings of a single basin theory, including a counterexample to the linear mode connectivity hypothesis.

Iterative Linear Quadratic Optimization for Nonlinear Control: Differentiable Programming Algorithmic Templates

We present the implementation of nonlinear control algorithms based on linear and quadratic approximations of the objective from a functional viewpoint. We present a gradient descent, a Gauss-Newton method, a Newton method, differential dynamic programming approaches with linear quadratic or quadratic approximations, various line-search strategies, and regularized variants of these algorithms. We derive the computational complexities of all algorithms in a differentiable programming framework and present sufficient optimality conditions. We compare the algorithms on several benchmarks, such as autonomous car racing using a bicycle model of a car. The algorithms are coded in a differentiable programming language in a publicly available package.

Electrostatic Brakes Enable Individual Joint Control of Underactuated, Highly Articulated Robots

Highly articulated organisms serve as blueprints for incredibly dexterous mechanisms, but building similarly capable robotic counterparts has been hindered by the difficulties of developing electromechanical actuators with both the high strength and compactness of biological muscle. We develop a stackable electrostatic brake that has comparable specific tension and weight to that of muscles and integrate it into a robotic joint. Compared to electromechanical motors, our brake-equipped joint is four times lighter and one thousand times more power efficient while exerting similar holding torques. Our joint design enables a ten degree-of-freedom robot equipped with only one motor to manipulate multiple objects simultaneously. We also show that the use of brakes allows a two-fingered robot to perform in-hand re-positioning of an object 45% more quickly and with 53% lower positioning error than without brakes. Relative to fully actuated robots, our findings suggest that robots equipped with such electrostatic brakes will have lower weight, volume, and power consumption yet retain the ability to reach arbitrary joint configurations.

Balancing Efficiency and Comfort in Robot-Assisted Bite Transfer

Robot-assisted feeding in household environments is challenging because it requires robots to generate trajectories that effectively bring food items of varying shapes and sizes into the mouth while making sure the user is comfortable. Our key insight is that in order to solve this challenge, robots must balance the efficiency of feeding a food item with the comfort of each individual bite. We formalize comfort and efficiency as heuristics to incorporate in motion planning. We present an approach based on heuristics-guided bi-directional Rapidly-exploring Random Trees (h-BiRRT) that selects bite transfer trajectories of arbitrary food item geometries and shapes using our developed bite efficiency and comfort heuristics and a learned constraint model. Real-robot evaluations show that optimizing both comfort and efficiency significantly outperforms a fixed-pose based method, and users preferred our method significantly more than that of a method that maximizes only user comfort. Videos and Appendices are found on our website: https://sites.google.com/view/comfortbitetransfer-icra22/home.

Leveraging Experience in Lazy Search

Lazy graph search algorithms are efficient at solving motion planning problems where edge evaluation is the computational bottleneck. These algorithms work by lazily computing the shortest potentially feasible path, evaluating edges along that path, and repeating until a feasible path is found. The order in which edges are selected is critical to minimizing the total number of edge evaluations: a good edge selector chooses edges that are not only likely to be invalid, but also eliminates future paths from consideration. We wish to learn such a selector by leveraging prior experience. We formulate this problem as a Markov Decision Process (MDP) on the state of the search problem. While solving this large MDP is generally intractable, we show that we can compute oracular selectors that can solve the MDP during training. With access to such oracles, we use imitation learning to find effective policies. If new search problems are sufficiently similar to problems solved during training, the learned policy will choose a good edge evaluation ordering and solve the motion planning problem quickly. We evaluate our algorithms on a wide range of 2D and 7D problems and show that the learned selector outperforms baseline commonly used heuristics. We further provide a novel theoretical analysis of lazy search in a Bayesian framework as well as regret guarantees on our imitation learning based approach to motion planning.

Lazy Lifelong Planning for Efficient Replanning in Graphs with Expensive Edge Evaluation

We present an incremental search algorithm, called Lifelong-GLS, which combines the vertex efficiency of Lifelong Planning A* (LPA*) and the edge efficiency of Generalized Lazy Search (GLS) for efficient replanning on dynamic graphs where edge evaluation is expensive. We use a lazily evaluated LPA* to repair the cost-to-come inconsistencies of the relevant region of the current search tree based on the previous search results, and then we restrict the expensive edge evaluations only to the current shortest subpath as in the GLS framework. The proposed algorithm is complete and correct in finding the optimal solution in the current graph, if one exists. We also show that the search returns a bounded suboptimal solution, if an inflated heuristic edge weight is used and the tree repairing propagation is truncated early for faster search. Finally, we show the efficiency of the proposed algorithm compared to the standard LPA* and the GLS algorithms over consecutive search episodes in a dynamic environment. For each search, the proposed algorithm reduces the edge evaluations by a significant amount compared to the LPA*. Both the number of vertex expansions and the number of edge evaluations are reduced substantially compared to GLS, as the proposed algorithm utilizes previous search results to facilitate the new search.

Faster Policy Learning with Continuous-Time Gradients

We study the estimation of policy gradients for continuous-time systems with known dynamics. By reframing policy learning in continuous-time, we show that it is possible construct a more efficient and accurate gradient estimator. The standard back-propagation through time estimator (BPTT) computes exact gradients for a crude discretization of the continuous-time system. In contrast, we approximate continuous-time gradients in the original system. With the explicit goal of estimating continuous-time gradients, we are able to discretize adaptively and construct a more efficient policy gradient estimator which we call the Continuous-Time Policy Gradient (CTPG). We show that replacing BPTT policy gradients with more efficient CTPG estimates results in faster and more robust learning in a variety of control tasks and simulators.

Grasping with Chopsticks: Combating Covariate Shift in Model-free Imitation Learning for Fine Manipulation

Billions of people use chopsticks, a simple yet versatile tool, for fine manipulation of everyday objects. The small, curved, and slippery tips of chopsticks pose a challenge for picking up small objects, making them a suitably complex test case. This paper leverages human demonstrations to develop an autonomous chopsticks-equipped robotic manipulator. Due to the lack of accurate models for fine manipulation, we explore model-free imitation learning, which traditionally suffers from the covariate shift phenomenon that causes poor generalization. We propose two approaches to reduce covariate shift, neither of which requires access to an interactive expert or a model, unlike previous approaches. First, we alleviate single-step prediction errors by applying an invariant operator to increase the data support at critical steps for grasping. Second, we generate synthetic corrective labels by adding bounded noise and combining parametric and non-parametric methods to prevent error accumulation. We demonstrate our methods on a real chopstick-equipped robot that we built, and observe the agent's success rate increase from 37.3% to 80%, which is comparable to the human expert performance of 82.6%.

Amodal 3D Reconstruction for Robotic Manipulation via Stability and Connectivity

Learning-based 3D object reconstruction enables single- or few-shot estimation of 3D object models. For robotics, this holds the potential to allow model-based methods to rapidly adapt to novel objects and scenes. Existing 3D reconstruction techniques optimize for visual reconstruction fidelity, typically measured by chamfer distance or voxel IOU. We find that when applied to realistic, cluttered robotics environments, these systems produce reconstructions with low physical realism, resulting in poor task performance when used for model-based control. We propose ARM, an amodal 3D reconstruction system that introduces (1) a stability prior over object shapes, (2) a connectivity prior, and (3) a multi-channel input representation that allows for reasoning over relationships between groups of objects. By using these priors over the physical properties of objects, our system improves reconstruction quality not just by standard visual metrics, but also performance of model-based control on a variety of robotics manipulation tasks in challenging, cluttered environments. Code is available at github.com/wagnew3/ARM.