Miguel Zamora

Advancing robotic manipulation of deformable objects can enable automation of repetitive tasks across multiple industries, from food processing to textiles and healthcare. Yet robots struggle with the high dimensionality of deformable objects and their complex dynamics. While data-driven methods have shown potential for solving manipulation tasks, their application in the domain of deformable objects has been constrained by the lack of data. To address this, we propose PokeFlex, a pilot dataset featuring real-world 3D mesh data of actively deformed objects, together with the corresponding forces applied by a robotic arm, using a simple poking strategy. Deformations are captured with a professional volumetric capture system that allows for complete 360-degree reconstruction. The PokeFlex dataset consists of five deformable objects with varying stiffness and shapes. Additionally, we leverage the PokeFlex dataset to train a vision model for online 3D mesh reconstruction from a single image and a template mesh. We refer readers to the supplementary material for demos and examples of our dataset.

Obtaining real world data for robotics tasks is harder than for other modalities such as vision and text. The data that is currently available for robot learning is mostly set in static scenes, and deals with a single robot only. Dealing with multiple robots comes with additional difficulties compared to single robot settings: the motion planning for multiple agents needs to take into account the movement of the other robots, and task planning needs to consider to which robot a task is assigned to, in addition to when a task should be done. In this work, we present TAPAS, a simulated dataset containing task and motion plans for multiple robots acting asynchronously in the same workspace and modifying the same environment. We consider prehensile manipulation in this dataset, and focus on various pick and place tasks. We demonstrate that training using this data for predicting makespan of a task sequence enables speeding up finding low makespan sequences by ranking sequences before computing the full motion plan.

A variety of control tasks such as inverse kinematics (IK), trajectory optimization (TO), and model predictive control (MPC) are commonly formulated as energy minimization problems. Numerical solutions to such problems are well-established. However, these are often too slow to be used directly in real-time applications. The alternative is to learn solution manifolds for control problems in an offline stage. Although this distillation process can be trivially formulated as a behavioral cloning (BC) problem, our experiments highlight a number of significant shortcomings arising due to incompatible local minima, interpolation artifacts, and insufficient coverage of the state space. In this paper, we propose an alternative to BC that is efficient and numerically robust. We formulate the learning of solution manifolds as a minimization of the energy terms of a control objective integrated over the space of problems of interest. We minimize this energy integral with a novel method that combines Monte Carlo-inspired adaptive sampling strategies with the derivatives used to solve individual instances of the control task. We evaluate the performance of our formulation on a series of robotic control problems of increasing complexity, and we highlight its benefits through comparisons against traditional methods such as behavioral cloning and Dataset aggregation (Dagger).

Current reinforcement learning (RL) methods use simulation models as simple black-box oracles. In this paper, with the goal of improving the performance exhibited by RL algorithms, we explore a systematic way of leveraging the additional information provided by an emerging class of differentiable simulators. Building on concepts established by Deterministic Policy Gradients (DPG) methods, the neural network policies learned with our approach represent deterministic actions. In a departure from standard methodologies, however, learning these policies does not hinge on approximations of the value function that must be learned concurrently in an actor-critic fashion. Instead, we exploit differentiable simulators to directly compute the analytic gradient of a policy’s value function with respect to the actions it outputs. This, in turn, allows us to efficiently perform locally optimal policy improvement iterations. Compared against other state-of-the-art RL methods, we show that with minimal hyper-parameter tuning our approach consistently leads to better asymptotic behavior across a set of payload manipulation tasks that demand a high degree of accuracy and precision.