Unsupervised Deformable Ultrasound Image Registration and Its Application for Vessel Segmentation

This paper presents a deep-learning model for deformable registration of ultrasound images at online rates, which we call U-RAFT. As its name suggests, U-RAFT is based on RAFT, a convolutional neural network for estimating optical flow. U-RAFT, however, can be trained in an unsupervised manner and can generate synthetic images for training vessel segmentation models. We propose and compare the registration quality of different loss functions for training U-RAFT. We also show how our approach, together with a robot performing force-controlled scans, can be used to generate synthetic deformed images to significantly expand the size of a femoral vessel segmentation training dataset without the need for additional manual labeling. We validate our approach on both a silicone human tissue phantom as well as on in-vivo porcine images. We show that U-RAFT generates synthetic ultrasound images with 98% and 81% structural similarity index measure (SSIM) to the real ultrasound images for the phantom and porcine datasets, respectively. We also demonstrate that synthetic deformed images from U-RAFT can be used as a data augmentation technique for vessel segmentation models to improve intersection-over-union (IoU) segmentation performance

Gait design for limbless obstacle aided locomotion using geometric mechanics

Limbless robots have the potential to maneuver through cluttered environments that conventional robots cannot traverse. As illustrated in their biological counterparts such as snakes and nematodes, limbless locomotors can benefit from interactions with obstacles, yet such obstacle-aided locomotion (OAL) requires properly coordinated high-level self-deformation patterns (gait templates) as well as low-level body adaptation to environments. Most prior work on OAL utilized stereotyped traveling-wave gait templates and relied on local body deformations (e.g., passive body mechanics or decentralized controller parameter adaptation based on force feedback) for obstacle navigation, while gait template design for OAL remains less studied. In this paper, we explore novel gait templates for OAL based on tools derived from geometric mechanics (GM), which thus far has been limited to homogeneous environments. Here, we expand the scope of GM to obstacle-rich environments. Specifically, we establish a model that maps the presence of an obstacle to directional constraints in optimization. In doing so, we identify novel gait templates suitable for sparsely and densely distributed obstacle-rich environments respectively. Open-loop robophysical experiments verify the effectiveness of our identified OAL gaits in obstacle-rich environments. We posit that when such OAL gait templates are augmented with appropriate sensing and feedback controls, limbless locomotors will gain robust function in obstacle rich environments.

Reslicing Ultrasound Images for Data Augmentation and Vessel Reconstruction

Robot-guided catheter insertion has the potential to deliver urgent medical care in situations where medical personnel are unavailable. However, this technique requires accurate and reliable segmentation of anatomical landmarks in the body. For the ultrasound imaging modality, obtaining large amounts of training data for a segmentation model is time-consuming and expensive. This paper introduces RESUS (RESlicing of UltraSound Images), a weak supervision data augmentation technique for ultrasound images based on slicing reconstructed 3D volumes from tracked 2D images. This technique allows us to generate views which cannot be easily obtained in vivo due to physical constraints of ultrasound imaging, and use these augmented ultrasound images to train a semantic segmentation model. We demonstrate that RESUS achieves statistically significant improvement over training with non-augmented images and highlight qualitative improvements through vessel reconstruction.

Analysis of Interior Rubble Void Spaces at Champlain Towers South Collapse

The 2021 Champlain Towers South Condominiums collapse in Surfside, Florida, resulted 98 deaths. Nine people are thought to have survived the initial collapse, and might have been rescued if rescue workers could have located them. Perhaps, if rescue workers had been able to use robots to search the interior of the rubble pile, outcomes might have been better. An improved understanding of the environment in which a robot would have to operate to be able to search the interior of a rubble pile would help roboticists develop better suited robotic platforms and control strategies. To this end, this work offers an approach to characterize and visualize the interior of a rubble pile and conduct a preliminary analysis of the occurrence of voids. Specifically, the analysis makes opportunistic use of four days of aerial imagery gathered from responders at Surfside to create a 3D volumetric aggregated model of the collapse in order to identify and characterize void spaces in the interior of the rubble. The preliminary results confirm expectations of small number and scale of these interior voids. The results can inform better selection and control of existing robots for disaster response, aid in determining the design specifications (specifically scale and form factor), and improve control of future robotic platforms developed for search operations in rubble.

Enhanced Multi-Objective A* with Partial Expansion

The Multi-Objective Shortest Path Problem, typically posed on a graph, determines a set of paths from a start vertex to a destination vertex while optimizing multiple objectives. In general, there does not exist a single solution path that can simultaneously optimize all the objectives and the problem thus seeks to find a set of so-called Pareto-optimal solutions. To address this problem, several Multi-Objective A* (MOA*) algorithms were recently developed to quickly compute solutions with quality guarantees. However, these MOA* algorithms often suffer from high memory usage, especially when the branching factor (i.e., the number of neighbors of any vertex) of the graph is large. This work thus aims at reducing the high memory consumption of MOA* with little increase in the runtime. In this paper, we first extend the notion of "partial expansion" (PE) from single-objective to multi-objective and then fuse this new PE technique with EMOA*, a recent runtime efficient MOA* algorithm. Furthermore, the resulting algorithm PE-EMOA* can balance between runtime and memory efficiency by tuning a user-defined hyper-parameter.

Learning Modular Robot Locomotion from Demonstrations

Modular robots can be reconfigured to create a variety of designs from a small set of components. But constructing a robot's hardware on its own is not enough -- each robot needs a controller. One could create controllers for some designs individually, but developing policies for additional designs can be time consuming. This work presents a method that uses demonstrations from one set of designs to accelerate policy learning for additional designs. We leverage a learning framework in which a graph neural network is made up of modular components, each component corresponds to a type of module (e.g., a leg, wheel, or body) and these components can be recombined to learn from multiple designs at once. In this paper we develop a combined reinforcement and imitation learning algorithm. Our method is novel because the policy is optimized to both maximize a reward for one design, and simultaneously imitate demonstrations from different designs, within one objective function. We show that when the modular policy is optimized with this combined objective, demonstrations from one set of designs influence how the policy behaves on a different design, decreasing the number of training iterations needed.

Learning Modular Robot Visual-motor Locomotion Policies

Control policy learning for modular robot locomotion has previously been limited to proprioceptive feedback and flat terrain. This paper develops policies for modular systems with vision traversing more challenging environments. These modular robots can be reconfigured to form many different designs, where each design needs a controller to function. Though one could create a policy for individual designs and environments, such an approach is not scalable given the wide range of potential designs and environments. To address this challenge, we create a visual-motor policy that can generalize to both new designs and environments. The policy itself is modular, in that it is divided into components, each of which corresponds to a type of module (e.g., a leg, wheel, or body). The policy components can be recombined during training to learn to control multiple designs. We develop a deep reinforcement learning algorithm where visual observations are input to a modular policy interacting with multiple environments at once. We apply this algorithm to train robots with combinations of legs and wheels, then demonstrate the policy controlling real robots climbing stairs and curbs.

Mathematical Justification of Hard Negative Mining via Isometric Approximation Theorem

In deep metric learning, the Triplet Loss has emerged as a popular method to learn many computer vision and natural language processing tasks such as facial recognition, object detection, and visual-semantic embeddings. One issue that plagues the Triplet Loss is network collapse, an undesirable phenomenon where the network projects the embeddings of all data onto a single point. Researchers predominately solve this problem by using triplet mining strategies. While hard negative mining is the most effective of these strategies, existing formulations lack strong theoretical justification for their empirical success. In this paper, we utilize the mathematical theory of isometric approximation to show an equivalence between the Triplet Loss sampled by hard negative mining and an optimization problem that minimizes a Hausdorff-like distance between the neural network and its ideal counterpart function. This provides the theoretical justifications for hard negative mining's empirical efficacy. In addition, our novel application of the isometric approximation theorem provides the groundwork for future forms of hard negative mining that avoid network collapse. Our theory can also be extended to analyze other Euclidean space-based metric learning methods like Ladder Loss or Contrastive Learning.

Torque-Limited Manipulation Planning through Contact by Interleaving Graph Search and Trajectory Optimization

Robots often have to perform manipulation tasks in close proximity to people. As such, it is desirable to use a robot arm that has limited joint torques so as to not injure the nearby person. Unfortunately, these limited torques then limit the payload capability of the arm. By using contact with the environment, robots can expand their reachable workspace that, otherwise, would be inaccessible due to exceeding actuator torque limits. We adapt our recently developed INSAT algorithm \cite{insat} to tackle the problem of torque-limited whole arm manipulation planning through contact. INSAT requires no prior over contact mode sequence and no initial template or seed for trajectory optimization. INSAT achieves this by interleaving graph search to explore the manipulator joint configuration space with incremental trajectory optimizations seeded by neighborhood solutions to find a dynamically feasible trajectory through contact. We demonstrate our results on a variety of manipulators and scenarios in simulation. We also experimentally show our planner exploiting robot-environment contact for the pick and place of a payload using a Kinova Gen3 robot. In comparison to the same trajectory running in free space, we experimentally show that the utilization of bracing contacts reduces the overall torque required to execute the trajectory.

GLSO: Grammar-guided Latent Space Optimization for Sample-efficient Robot Design Automation

Robots have been used in all sorts of automation, and yet the design of robots remains mainly a manual task. We seek to provide design tools to automate the design of robots themselves. An important challenge in robot design automation is the large and complex design search space which grows exponentially with the number of components, making optimization difficult and sample inefficient. In this work, we present Grammar-guided Latent Space Optimization (GLSO), a framework that transforms design automation into a low-dimensional continuous optimization problem by training a graph variational autoencoder (VAE) to learn a mapping between the graph-structured design space and a continuous latent space. This transformation allows optimization to be conducted in a continuous latent space, where sample efficiency can be significantly boosted by applying algorithms such as Bayesian Optimization. GLSO guides training of the VAE using graph grammar rules and robot world space features, such that the learned latent space focus on valid robots and is easier for the optimization algorithm to explore. Importantly, the trained VAE can be reused to search for designs specialized to multiple different tasks without retraining. We evaluate GLSO by designing robots for a set of locomotion tasks in simulation, and demonstrate that our method outperforms related state-of-the-art robot design automation methods.