In this article, we address the task and safety performance of data-driven model predictive controllers (DD-MPC) for systems with complex dynamics, i.e., temporally or spatially varying dynamics that may also be discontinuous. The three challenges we focus on are the accuracy of learned models, the receding horizon-induced myopic predictions of DD-MPC, and the active encouragement of safety. To learn accurate models for DD-MPC, we cautiously, yet effectively, explore the dynamical system with rapidly exploring random trees (RRT) to collect a uniform distribution of samples in the state-input space and overcome the common distribution shift in model learning. The learned model is further used to construct an RRT tree that estimates how close the model's predictions are to the desired target. This information is used in the cost function of the DD-MPC to minimize the short-sighted effect of its receding horizon nature. To promote safety, we approximate sets of safe states using demonstrations of exclusively safe trajectories, i.e., without unsafe examples, and encourage the controller to generate trajectories close to the sets. As a running example, we use a broken version of an inverted pendulum where the friction abruptly changes in certain regions. Furthermore, we showcase the adaptation of our method to a real-world robotic application with complex dynamics: robotic food-cutting. Our results show that our proposed control framework effectively avoids unsafe states with higher success rates than baseline controllers that employ models from controlled demonstrations and even random actions.

Deep Neural Networks (NNs) have been widely utilized in contact-rich manipulation tasks to model the complicated contact dynamics. However, NN-based models are often difficult to decipher which can lead to seemingly inexplicable behaviors and unidentifiable failure cases. In this work, we address the interpretability of NN-based models by introducing the kinodynamic images. We propose a methodology that creates images from kinematic and dynamic data of contact-rich manipulation tasks. By using images as the state representation, we enable the application of interpretability modules that were previously limited to vision-based tasks. We use this representation to train a Convolutional Neural Network (CNN) and we extract interpretations with Grad-CAM to produce visual explanations. Our method is versatile and can be applied to any classification problem in manipulation tasks to visually interpret which parts of the input drive the model’s decisions and distinguish its failure modes, regardless of the features used. Our experiments demonstrate that our method enables detailed visual inspections of sequences in a task, and high-level evaluations of a model’s behavior.

Interaction dynamics are difficult to model analytically, making data-driven controllers preferable for contact-rich manipulation tasks. In this work, we approximate the intricate dynamics of food-cutting with a Long Short-Term Memory (LSTM) model to apply a Model Predictive Controller (MPC). We propose a problem formulation that allows velocity-controlled robots to learn the interaction dynamics and tackle the difficulty of multi-step predictions by training the model with a horizon curriculum. We experimentally demonstrate that our approach leads to good predictive performance that scales for longer prediction horizons, generalizes to unseen object classes and results in controller behaviors with an understanding of the cutting dynamics.

In this paper, we address the safety of data-driven control for contact-rich manipulation. We propose to restrict the controller's action space to keep the system in a set of safe states. In the absence of an analytical model, we show how Gaussian Processes (GP) can be used to approximate safe sets. We disable inputs for which the predicted states are likely to be unsafe using the GP. Furthermore, we show how locally designed feedback controllers can be used to improve the execution precision in the presence of modelling errors. We demonstrate the benefits of our method on a pushing task with a variety of dynamics, by using known and unknown surfaces and different object loads. Our results illustrate that the proposed approach significantly improves the performance and safety of the baseline controller.

— Identification of textile properties is an important milestone toward advanced robotic manipulation tasks that consider interaction with clothing items such as assisted dressing, laundry folding, automated sewing, textile recycling and reusing. Despite the abundance of work considering this class of deformable objects, many open problems remain. These relate to the choice and modelling of the sensory feedback as well as the control and planning of the interaction and manipulation strategies. Most importantly, there is no structured approach for studying and assessing different approaches that may bridge the gap between the robotics community and textile production industry. To this end, we outline a textile taxonomy considering fiber types and production methods, commonly used in textile industry. We devise datasets according to the taxonomy, and study how robotic actions, such as pulling and twisting of the textile samples, can be used for the classification. We also provide important insights from the perspective of visualization and interpretability of the gathered data.

In this paper, we address the safety of data-driven control for contact-rich manipulation. We propose to restrict the controller’s action space to keep the system in a set of safe states. In the absence of an analytical model, we show how Gaussian Processes (GP) can be used to approximate safe sets. We disable inputs for which the predicted states are likely to be unsafe using the GP. Furthermore, we show how locally designed feedback controllers can be used to improve the execution precision in the presence of modelling errors. We demonstrate the benefits of our method on a pushing task with a variety of dynamics, by using known and unknown surfaces and different object loads. Our results illustrate that the proposed approach significantly improves the performance and safety of the baseline controller.

Modelling of contact-rich tasks is challenging and cannot be entirely solved using classical control approaches due to the difficulty of constructing an analytic description of the contact dynamics. Additionally, in a manipulation task like food-cutting, purely learning-based methods such as Reinforcement Learning, require either a vast amount of data that is expensive to collect on a real robot, or a highly realistic simulation environment, which is currently not available. This paper presents a data-driven control approach that employs a recurrent neural network to model the dynamics for a Model Predictive Controller. We build upon earlier work limited to torque-controlled robots and redefine it for velocity controlled ones. We incorporate force/torque sensor measurements, reformulate and further extend the control problem formulation. We evaluate the performance on objects used for training, as well as on unknown objects, by means of the cutting rates achieved and demonstrate that the method can efficiently treat different cases with only one dynamic model. Finally we investigate the behavior of the system during force-critical instances of cutting and illustrate its adaptive behavior in difficult cases.

In this paper, we address the safety and efficiency of data-driven model predictive controllers (DD-MPC) for systems with complex dynamics. First, we utilize safe exploration of dynamical systems to learn an accurate model for the DD-MPC. During training, we use rapidly exploring random trees (RRT) to collect a uniform distribution of data points in the state-input space and overcome the common distribution shift in model learning. This model is also used to construct a tree offline, which at test time is used in the cost function to provide an estimate of the predicted states' distance to the target. Additionally, we show how safe sets can be approximated using demonstrations of exclusively safe trajectories, i.e. positive examples. During test time, the distances of the predicted trajectories to the safe set are used as a cost term to encourage safe inputs. We use a \emph{broken} version of the inverted pendulum problem where the friction abruptly changes in certain regions as a running example. Our results show that the proposed exploration algorithm and the two proposed cost terms lead to a controller that can effectively avoid unsafe states and displays higher success rates than the baseline controllers with models from controlled demonstrations and even random actions.