Task-Based Design and Policy Co-Optimization for Tendon-driven Underactuated Kinematic Chains

Underactuated manipulators reduce the number of bulky motors, thereby enabling compact and mechanically robust designs. However, fewer actuators than joints means that the manipulator can only access a specific manifold within the joint space, which is particular to a given hardware configuration and can be low-dimensional and/or discontinuous. Determining an appropriate set of hardware parameters for this class of mechanisms, therefore, is difficult - even for traditional task-based co-optimization methods. In this paper, our goal is to implement a task-based design and policy co-optimization method for underactuated, tendon-driven manipulators. We first formulate a general model for an underactuated, tendon-driven transmission. We then use this model to co-optimize a three-link, two-actuator kinematic chain using reinforcement learning. We demonstrate that our optimized tendon transmission and control policy can be transferred reliably to physical hardware with real-world reaching experiments.

MORPH: Design Co-optimization with Reinforcement Learning via a Differentiable Hardware Model Proxy

We introduce MORPH, a method for co-optimization of hardware design parameters and control policies in simulation using reinforcement learning. Like most co-optimization methods, MORPH relies on a model of the hardware being optimized, usually simulated based on the laws of physics. However, such a model is often difficult to integrate into an effective optimization routine. To address this, we introduce a proxy hardware model, which is always differentiable and enables efficient co-optimization alongside a long-horizon control policy using RL. MORPH is designed to ensure that the optimized hardware proxy remains as close as possible to its realistic counterpart, while still enabling task completion. We demonstrate our approach on simulated 2D reaching and 3D multi-fingered manipulation tasks.

Pick2Place: Task-aware 6DoF Grasp Estimation via Object-Centric Perspective Affordance

The choice of a grasp plays a critical role in the success of downstream manipulation tasks. Consider a task of placing an object in a cluttered scene; the majority of possible grasps may not be suitable for the desired placement. In this paper, we study the synergy between the picking and placing of an object in a cluttered scene to develop an algorithm for task-aware grasp estimation. We present an object-centric action space that encodes the relationship between the geometry of the placement scene and the object to be placed in order to provide placement affordance maps directly from perspective views of the placement scene. This action space enables the computation of a one-to-one mapping between the placement and picking actions allowing the robot to generate a diverse set of pick-and-place proposals and to optimize for a grasp under other task constraints such as robot kinematics and collision avoidance. With experiments both in simulation and on a real robot we demonstrate that with our method, the robot is able to successfully complete the task of placement-aware grasping with over 89% accuracy in such a way that generalizes to novel objects and scenes.

Decision Making for Human-in-the-loop Robotic Agents via Uncertainty-Aware Reinforcement Learning

In a Human-in-the-Loop paradigm, a robotic agent is able to act mostly autonomously in solving a task, but can request help from an external expert when needed. However, knowing when to request such assistance is critical: too few requests can lead to the robot making mistakes, but too many requests can overload the expert. In this paper, we present a Reinforcement Learning based approach to this problem, where a semi-autonomous agent asks for external assistance when it has low confidence in the eventual success of the task. The confidence level is computed by estimating the variance of the return from the current state. We show that this estimate can be iteratively improved during training using a Bellman-like recursion. On discrete navigation problems with both fully- and partially-observable state information, we show that our method makes effective use of a limited budget of expert calls at run-time, despite having no access to the expert at training time.

Discovering Synergies for Robot Manipulation with Multi-Task Reinforcement Learning

Controlling robotic manipulators with high-dimensional action spaces for dexterous tasks is a challenging problem. Inspired by human manipulation, researchers have studied generating and using postural synergies for robot hands to accomplish manipulation tasks, leveraging the lower dimensional nature of synergistic action spaces. However, many of these works require pre-collected data from an existing controller in order to derive such a subspace by means of dimensionality reduction. In this paper, we present a framework that simultaneously discovers a synergy space and a multi-task policy that operates on this low-dimensional action space to accomplish diverse manipulation tasks. We demonstrate that our end-to-end method is able to perform multiple tasks using few synergies, and outperforms sequential methods that apply dimensionality reduction to independently collected data. We also show that deriving synergies using multiple tasks can lead to a subspace that enables robots to efficiently learn new manipulation tasks and interactions with new objects.

UMPNet: Universal Manipulation Policy Network for Articulated Objects

We introduce the Universal Manipulation Policy Network (UMPNet) -- a single image-based policy network that infers closed-loop action sequences for manipulating arbitrary articulated objects. To infer a wide range of action trajectories, the policy supports 6DoF action representation and varying trajectory length. To handle a diverse set of objects, the policy learns from objects with different articulation structures and generalizes to unseen objects or categories. The policy is trained with self-guided exploration without any human demonstrations, scripted policy, or pre-defined goal conditions. To support effective multi-step interaction, we introduce a novel Arrow-of-Time action attribute that indicates whether an action will change the object state back to the past or forward into the future. With the Arrow-of-Time inference at each interaction step, the learned policy is able to select actions that consistently lead towards or away from a given state, thereby, enabling both effective state exploration and goal-conditioned manipulation. Video is available at https://youtu.be/KqlvcL9RqKM

Learning 3D Dynamic Scene Representations for Robot Manipulation

3D scene representation for robot manipulation should capture three key object properties: permanency -- objects that become occluded over time continue to exist; amodal completeness -- objects have 3D occupancy, even if only partial observations are available; spatiotemporal continuity -- the movement of each object is continuous over space and time. In this paper, we introduce 3D Dynamic Scene Representation (DSR), a 3D volumetric scene representation that simultaneously discovers, tracks, reconstructs objects, and predicts their dynamics while capturing all three properties. We further propose DSR-Net, which learns to aggregate visual observations over multiple interactions to gradually build and refine DSR. Our model achieves state-of-the-art performance in modeling 3D scene dynamics with DSR on both simulated and real data. Combined with model predictive control, DSR-Net enables accurate planning in downstream robotic manipulation tasks such as planar pushing. Video is available at https://youtu.be/GQjYG3nQJ80.

Hardware as Policy: Mechanical and Computational Co-Optimization using Deep Reinforcement Learning

Deep Reinforcement Learning (RL) has shown great success in learning complex control policies for a variety of applications in robotics. However, in most such cases, the hardware of the robot has been considered immutable, modeled as part of the environment. In this study, we explore the problem of learning hardware and control parameters together in a unified RL framework. To achieve this, we propose to model aspects of the robot's hardware as a "mechanical policy", analogous to and optimized jointly with its computational counterpart. We show that, by modeling such mechanical policies as auto-differentiable computational graphs, the ensuing optimization problem can be solved efficiently by gradient-based algorithms from the Policy Optimization family. We present two such design examples: a toy mass-spring problem, and a real-world problem of designing an underactuated hand. We compare our method against traditional co-optimization approaches, and also demonstrate its effectiveness by building a physical prototype based on the learned hardware parameters.

SQUIRL: Robust and Efficient Learning from Video Demonstration of Long-Horizon Robotic Manipulation Tasks

Recent advances in deep reinforcement learning (RL) have demonstrated its potential to learn complex robotic manipulation tasks. However, RL still requires the robot to collect a large amount of real-world experience. To address this problem, recent works have proposed learning from expert demonstrations (LfD), particularly via inverse reinforcement learning (IRL), given its ability to achieve robust performance with only a small number of expert demonstrations. Nevertheless, deploying IRL on real robots is still challenging due to the large number of robot experiences it requires. This paper aims to address this scalability challenge with a robust, sample-efficient, and general meta-IRL algorithm, SQUIRL, that performs a new but related long-horizon task robustly given only a single video demonstration. First, this algorithm bootstraps the learning of a task encoder and a task-conditioned policy using behavioral cloning (BC). It then collects real-robot experiences and bypasses reward learning by directly recovering a Q-function from the combined robot and expert trajectories. Next, this algorithm uses the Q-function to re-evaluate all cumulative experiences collected by the robot to improve the policy quickly. In the end, the policy performs more robustly (90%+ success) than BC on new tasks while requiring no trial-and-errors at test time. Finally, our real-robot and simulated experiments demonstrate our algorithm's generality across different state spaces, action spaces, and vision-based manipulation tasks, e.g., pick-pour-place and pick-carry-drop.

Meta-World: A Benchmark and Evaluation for Multi-Task and Meta Reinforcement Learning

Meta-reinforcement learning algorithms can enable robots to acquire new skills much more quickly, by leveraging prior experience to learn how to learn. However, much of the current research on meta-reinforcement learning focuses on task distributions that are very narrow. For example, a commonly used meta-reinforcement learning benchmark uses different running velocities for a simulated robot as different tasks. When policies are meta-trained on such narrow task distributions, they cannot possibly generalize to more quickly acquire entirely new tasks. Therefore, if the aim of these methods is to enable faster acquisition of entirely new behaviors, we must evaluate them on task distributions that are sufficiently broad to enable generalization to new behaviors. In this paper, we propose an open-source simulated benchmark for meta-reinforcement learning and multi-task learning consisting of 50 distinct robotic manipulation tasks. Our aim is to make it possible to develop algorithms that generalize to accelerate the acquisition of entirely new, held-out tasks. We evaluate 6 state-of-the-art meta-reinforcement learning and multi-task learning algorithms on these tasks. Surprisingly, while each task and its variations (e.g., with different object positions) can be learned with reasonable success, these algorithms struggle to learn with multiple tasks at the same time, even with as few as ten distinct training tasks. Our analysis and open-source environments pave the way for future research in multi-task learning and meta-learning that can enable meaningful generalization, thereby unlocking the full potential of these methods.