Protective Policy Transfer

Being able to transfer existing skills to new situations is a key capability when training robots to operate in unpredictable real-world environments. A successful transfer algorithm should not only minimize the number of samples that the robot needs to collect in the new environment, but also prevent the robot from damaging itself or the surrounding environment during the transfer process. In this work, we introduce a policy transfer algorithm for adapting robot motor skills to novel scenarios while minimizing serious failures. Our algorithm trains two control policies in the training environment: a task policy that is optimized to complete the task of interest, and a protective policy that is dedicated to keep the robot from unsafe events (e.g. falling to the ground). To decide which policy to use during execution, we learn a safety estimator model in the training environment that estimates a continuous safety level of the robot. When used with a set of thresholds, the safety estimator becomes a classifier for switching between the protective policy and the task policy. We evaluate our approach on four simulated robot locomotion problems and a 2D navigation problem and show that our method can achieve successful transfer to notably different environments while taking the robot's safety into consideration.

Perspectives on Sim2Real Transfer for Robotics: A Summary of the R:SS 2020 Workshop

This report presents the debates, posters, and discussions of the Sim2Real workshop held in conjunction with the 2020 edition of the "Robotics: Science and System" conference. Twelve leaders of the field took competing debate positions on the definition, viability, and importance of transferring skills from simulation to the real world in the context of robotics problems. The debaters also joined a large panel discussion, answering audience questions and outlining the future of Sim2Real in robotics. Furthermore, we invited extended abstracts to this workshop which are summarized in this report. Based on the workshop, this report concludes with directions for practitioners exploiting this technology and for researchers further exploring open problems in this area.

COCOI: Contact-aware Online Context Inference for Generalizable Non-planar Pushing

General contact-rich manipulation problems are long-standing challenges in robotics due to the difficulty of understanding complicated contact physics. Deep reinforcement learning (RL) has shown great potential in solving robot manipulation tasks. However, existing RL policies have limited adaptability to environments with diverse dynamics properties, which is pivotal in solving many contact-rich manipulation tasks. In this work, we propose Contact-aware Online COntext Inference (COCOI), a deep RL method that encodes a context embedding of dynamics properties online using contact-rich interactions. We study this method based on a novel and challenging non-planar pushing task, where the robot uses a monocular camera image and wrist force torque sensor reading to push an object to a goal location while keeping it upright. We run extensive experiments to demonstrate the capability of COCOI in a wide range of settings and dynamics properties in simulation, and also in a sim-to-real transfer scenario on a real robot (Video: https://youtu.be/nrmJYksh1Kc)

Learning Human Search Behavior from Egocentric Visual Inputs

"Looking for things" is a mundane but critical task we repeatedly carry on in our daily life. We introduce a method to develop a human character capable of searching for a randomly located target object in a detailed 3D scene using its locomotion capability and egocentric vision perception represented as RGBD images. By depriving the privileged 3D information from the human character, it is forced to move and look around simultaneously to account for the restricted sensing capability, resulting in natural navigation and search behaviors. Our method consists of two components: 1) a search control policy based on an abstract character model, and 2) an online replanning control module for synthesizing detailed kinematic motion based on the trajectories planned by the search policy. We demonstrate that the combined techniques enable the character to effectively find often occluded household items in indoor environments. The same search policy can be applied to different full-body characters without the need for retraining. We evaluate our method quantitatively by testing it on randomly generated scenarios. Our work is a first step toward creating intelligent virtual agents with humanlike behaviors driven by onboard sensors, paving the road toward future robotic applications.

Policy Transfer via Kinematic Domain Randomization and Adaptation

Transferring reinforcement learning policies trained in physics simulation to the real hardware remains a challenge, known as the "sim-to-real" gap. Domain randomization is a simple yet effective technique to address dynamics discrepancies across source and target domains, but its success generally depends on heuristics and trial-and-error. In this work we investigate the impact of randomized parameter selection on policy transferability across different types of domain discrepancies. Contrary to common practice in which kinematic parameters are carefully measured while dynamic parameters are randomized, we found that virtually randomizing kinematic parameters (e.g., link lengths) during training in simulation generally outperforms dynamic randomization. Based on this finding, we introduce a new domain adaptation algorithm that utilizes simulated kinematic parameters variation. Our algorithm, Multi-Policy Bayesian Optimization, trains an ensemble of universal policies conditioned on virtual kinematic parameters and efficiently adapts to the target environment using a limited number of target domain rollouts. We showcase our findings on a simulated quadruped robot in five different target environments covering different aspects of domain discrepancies.

Learning Task-Agnostic Action Spaces for Movement Optimization

We propose a novel method for exploring the dynamics of physically based animated characters, and learning a task-agnostic action space that makes movement optimization easier. Like several previous papers, we parameterize actions as target states, and learn a short-horizon goal-conditioned low-level control policy that drives the agent's state towards the targets. Our novel contribution is that with our exploration data, we are able to learn the low-level policy in a generic manner and without any reference movement data. Trained once for each agent or simulation environment, the policy improves the efficiency of optimizing both trajectories and high-level policies across multiple tasks and optimization algorithms. We also contribute novel visualizations that show how using target states as actions makes optimized trajectories more robust to disturbances; this manifests as wider optima that are easy to find. Due to its simplicity and generality, our proposed approach should provide a building block that can improve a large variety of movement optimization methods and applications.

Bodies at Rest: 3D Human Pose and Shape Estimation from a Pressure Image using Synthetic Data

People spend a substantial part of their lives at rest in bed. 3D human pose and shape estimation for this activity would have numerous beneficial applications, yet line-of-sight perception is complicated by occlusion from bedding. Pressure sensing mats are a promising alternative, but training data is challenging to collect at scale. We describe a physics-based method that simulates human bodies at rest in a bed with a pressure sensing mat, and present PressurePose, a synthetic dataset with 206K pressure images with 3D human poses and shapes. We also present PressureNet, a deep learning model that estimates human pose and shape given a pressure image and gender. PressureNet incorporates a pressure map reconstruction (PMR) network that models pressure image generation to promote consistency between estimated 3D body models and pressure image input. In our evaluations, PressureNet performed well with real data from participants in diverse poses, even though it had only been trained with synthetic data. When we ablated the PMR network, performance dropped substantially.

Assistive Gym: A Physics Simulation Framework for Assistive Robotics

Autonomous robots have the potential to serve as versatile caregivers that improve quality of life for millions of people worldwide. Yet, conducting research in this area presents numerous challenges, including the risks of physical interaction between people and robots. Physics simulations have been used to optimize and train robots for physical assistance, but have typically focused on a single task. In this paper, we present Assistive Gym, an open source physics simulation framework for assistive robots that models multiple tasks. It includes six simulated environments in which a robotic manipulator can attempt to assist a person with activities of daily living (ADLs): itch scratching, drinking, feeding, body manipulation, dressing, and bathing. Assistive Gym models a person's physical capabilities and preferences for assistance, which are used to provide a reward function. We present baseline policies trained using reinforcement learning for four different commercial robots in the six environments. We demonstrate that modeling human motion results in better assistance and we compare the performance of different robots. Overall, we show that Assistive Gym is a promising tool for assistive robotics research.

Learning a Control Policy for Fall Prevention on an Assistive Walking Device

Fall prevention is one of the most important components in senior care. We present a technique to augment an assistive walking device with the ability to prevent falls. Given an existing walking device, our method develops a fall predictor and a recovery policy by utilizing the onboard sensors and actuators. The key component of our method is a robust human walking policy that models realistic human gait under a moderate level of perturbations. We use this human walking policy to provide training data for the fall predictor, as well as to teach the recovery policy how to best modify the person's gait when a fall is imminent. Our evaluation shows that the human walking policy generates walking sequences similar to those reported in biomechanics literature. Our experiments in simulation show that the augmented assistive device can indeed help recover balance from a variety of external perturbations. We also provide a quantitative method to evaluate the design choices for an assistive device.

Visualizing Movement Control Optimization Landscapes

A large body of animation research focuses on optimization of movement control, either as action sequences or policy parameters. However, as closed-form expressions of the objective functions are often not available, our understanding of the optimization problems is limited. Building on recent work on analyzing neural network training, we contribute novel visualizations of high-dimensional control optimization landscapes; this yields insights into why control optimization is hard and why common practices like early termination and spline-based action parameterizations make optimization easier. For example, our experiments show how trajectory optimization can become increasingly ill-conditioned with longer trajectories, but parameterizing control as partial target states - e.g., target angles converted to torques using a PD-controller - can act as an efficient preconditioner. Both our visualizations and quantitative empirical data also indicate that neural network policy optimization scales better than trajectory optimization for long planning horizons. Our work advances the understanding of movement optimization and our visualizations should also provide value in educational use.