Towards Agile Robotics: Creating Push-Off Skills for Dynamic Interactions
GHC 8102Abstract: Dynamic interactions play a fundamental role in human capabilities, enabling us to achieve a wide range of tasks such as moving heavy objects, manipulating our surroundings, and changing directions rapidly and safely. In contrast, most conventional robotic systems lack this level of agility and cannot perform dynamic interactions, limiting their potential in practical applications. [...]
Learning Safe Human-Robot Interactions for a Seamlessly Shared Airspace
NSH 3305Abstract: The growing need for fully autonomous aerial operations in shared spaces, necessitates the development of reliable agents capable of navigating safely and seamlessly alongside uncertain human agents. In response, we advocate endowing autonomous agents with the ability to predict human actions, comprehend and ground abstract rules in the action space, and embrace the uncertainty [...]
Generative Evolutionary Search with Diffusion Models for Trajectory Optimization
NSH 4305Abstract: Diffusion models excel at modeling complex and multimodal trajectory distributions for decision-making and control. Reward-gradient guided denoising has been recently proposed to generate trajectories that maximize both a differentiable reward function and the likelihood under the data distribution captured by a diffusion model. Reward-gradient guided denoising requires a differentiable reward function fitted to both [...]
Tartancalib: Iterative Wide-Angle Lens Calibration
GHC 8115Abstract: Mobile vision systems greatly benefit from the large field-of-view enabled by wide-angle lenses. Accurate and robust intrinsic calibration is a critical prerequisite for leveraging this property. Calibrating wide-angle lenses with current state-of-the-art techniques yields poor results due to extreme distortion at the edge. In this work, we present TartanCalib, an accurate and robust method [...]
Sample-Efficient Reinforcement Learning with applications in Nuclear Fusion
NSH 4305Abstract: In many practical applications of reinforcement learning (RL), it is expensive to observe state transitions from the environment. In the problem of plasma control for nuclear fusion, the motivating example of this thesis, determining the next state for a given state-action pair requires querying an expensive transition function which can lead to many hours [...]
[MSR Thesis Talk] Neural Implicit Representations for Medical Ultrasound Volumes and 3D Anatomy-specific Reconstructions
GHC 4405Abstract: Most Robotic Ultrasound Systems (RUSs) equipped with ultrasound-interpreting algorithms rely on building 3D reconstructions of the entire scanned region or specific anatomies. These 3D reconstructions are typically created via methods that compound or stack 2D tomographic ultrasound images using known poses of the ultrasound transducer with the latter requiring 2D or 3D segmentation. While fast, this class [...]