Student Talks
Continual Learning of Compositional Skills for Robust Robot Manipulation
Abstract: Real world robots need to continuously learn new manipulation tasks in a lifelong learning manner. These new tasks often share many sub-structures e.g. sub-tasks, controllers, preconditions, with previously learned tasks. To utilize these shared sub-structures, we explore a compositional and object-centric approach to learn manipulation tasks. The first part of this thesis focuses on [...]
Watch, Practice, Improve: Towards In-the-wild Manipulation
Abstract: The longstanding dream of many roboticists is to see robots perform diverse tasks in diverse environments. To build such a robot that can operate anywhere, many methods train on robotic interaction data. While these approaches have led to significant advances, they rely on heavily engineered setups or high amounts of supervision, neither of which [...]
Combining Physics-Based Light Transport and Neural Fields for Robust Inverse Rendering
Abstract: Inverse rendering — the process of recovering shape, material, and/or lighting of an object or environment from a set of images — is essential for applications in robotics and elsewhere, from AR/VR to perception on self-driving vehicles. While it is possible to perform inverse rendering from color images alone, it is often far easier [...]
Improving the Transparency of Agent Decision Making to Humans Using Demonstrations
Abstract: For intelligent agents (e.g. robots) to be seamlessly integrated into human society, humans must be able to understand their decision making. For example, the decision making of autonomous cars must be clear to the engineers certifying their safety, passengers riding them, and nearby drivers negotiating the road simultaneously. As an agent's decision making depends [...]
Robotic Climbing for Extreme Terrain Exploration
Abstract: Climbing robots can operate in steep and unstructured environments that are inaccessible to other ground robots, with applications ranging from the inspection of artificial structures on Earth to the exploration of natural terrain features throughout the solar system. Climbing robots for planetary exploration face many challenges to deployment, including mass restrictions, irregular surface features, [...]
Perception amidst interaction: spatial AI with vision and touch for robot manipulation
Abstract: Robots currently lack the cognition to replicate even a fraction of the tasks humans do, a trend summarized by Moravec's Paradox. Humans effortlessly combine their senses for everyday interactions—we can rummage through our pockets in search of our keys, and deftly insert them to unlock our front door. Before robots can demonstrate such dexterity, [...]
[MSR Thesis Talk] SplaTAM: Splat, Track & Map 3D Gaussians for Dense RGB-D SLAM
Abstract: Dense simultaneous localization and mapping (SLAM) is crucial for numerous robotic and augmented reality applications. However, current methods are often hampered by the non-volumetric or implicit way they represent a scene. This talk introduces SplaTAM, an approach that leverages explicit volumetric representations, i.e., 3D Gaussians, to enable high-fidelity reconstruction from a single unposed RGB-D [...]
Human Perception of Robot Failure and Explanation During a Pick-and-Place Task
Abstract: In recent years, researchers have extensively used non-verbal gestures, such as head and arm movements, to express the robot's intentions and capabilities to humans. Inspired by past research, we investigated how different explanation modalities can aid human understanding and perception of how robots communicate failures and provide explanations during block pick-and-place tasks. Through an in-person [...]
Learning Distributional Models for Relative Placement
Abstract: Relative placement tasks are an important category of tasks in which one object needs to be placed in a desired pose relative to another object. Previous work has shown success in learning relative placement tasks from just a small number of demonstrations, when using relational reasoning networks with geometric inductive biases. However, such methods fail [...]
Robust Body Exposure (RoBE): A Graph-based Dynamics Modeling Approach to Manipulating Blankets over People
Abstract: Robotic caregivers could potentially improve the quality of life of many who require physical assistance. However, in order to assist individuals who are lying in bed, robots must be capable of dealing with a significant obstacle: the blanket or sheet that will almost always cover the person's body. We propose a method for targeted [...]
Exploration for Continually Improving Robots
Abstract: General purpose robots should be able to perform arbitrary manipulation tasks, and get better at performing new ones as they obtain more experience. The current paradigm in robot learning involves imitation or simulation. Scaling these approaches to learn from more data for various tasks is bottle-necked by human labor required either in collecting demonstrations [...]
Sparse-view 3D in the Wild
Abstract: Reconstructing 3D scenes and objects from images alone has been a long-standing goal in computer vision. We have seen tremendous progress in recent years, capable of producing near photo-realistic renderings from any viewpoint. However, existing approaches generally rely on a large number of input images (typically 50-100) to compute camera poses and ensure view [...]
Deep 3D Geometric Reasoning for Robot Manipulation
Abstract: To solve general manipulation tasks in real-world environments, robots must be able to perceive and condition their manipulation policies on the 3D world. These agents will need to understand various common-sense spatial/geometric concepts about manipulation tasks: that local geometry can suggest potential manipulation strategies, that policies should be invariant across choice of reference frame, [...]
Probabilistic 3D Multi-Object Cooperative Tracking for Autonomous Driving via Differentiable Multi-Sensor Kalman Filter
This talk has been postponed […]
Towards diverse zero-shot manipulation via actualizing visual plans
Abstract: In this thesis, we seek to learn a generalizable goal-conditioned policy that enables zero-shot robot manipulation — interacting with unseen objects in novel scenes without test-time adaptation. Robots that can be reliably deployed out-of-the-box in new scenarios have the potential for helping humans in everyday tasks. Not requiring any test-time training through demonstrations or [...]
Deep Learning for Sensors: Development to Deployment
Abstract: Robots rely heavily on sensing to reason about physical interactions, and recent advancements in rapid prototyping, MEMS sensing, and machine learning have led to a plethora of sensing alternatives. However, few of these sensors have gained widespread use among roboticists. This thesis proposes a framework for incorporating sensors into a robot learning paradigm, from [...]
Offline Learning for Stochastic Multi-Agent Planning in Autonomous Driving
Abstract: Fully autonomous vehicles have the potential to greatly reduce vehicular accidents and revolutionize how people travel and how we transport goods. Many of the major challenges for autonomous driving systems emerge from the numerous traffic situations that require complex interactions with other agents. For the foreseeable future, autonomous vehicles will have to share the [...]
Transfer Learning via Temporal Contrastive Learning Inbox
Abstract: This thesis introduces a novel transfer learning framework for deep reinforcement learning. The approach automatically combines goal-conditioned policies with temporal contrastive learning to discover meaningful sub-goals. The approach involves pre-training a goal-conditioned agent, finetuning it on the target domain, and using contrastive learning to construct a planning graph that guides the agent via sub-goals. Experiments [...]
Towards Influence-Aware Safe Human-Robot Interaction
Abstract: In recent years, we have seen through recommender systems on social media how influential (and potentially harmful) algorithms can be in our lives, sometimes creating polarization and conspiracies that lead to unsafe behavior. Now that robots are also growing more common in the real world, we must be very careful to ensure that they [...]
Learning to Manipulate beyond Imitation
Abstract: Imitation learning has been a prevalent approach for teaching robots manipulation skills but still suffers from scalability and generalizability. In this talk, I'll argue for going beyond elementary behavioral imitation from human demonstrations. Instead, I'll present two key directions: 1) Creating Manipulation Controllers from Pre-Trained Representations, and 2) Representing Video Demonstrations with Parameterized Symbolic [...]
Improving Robot Capabilities Through Reconfigurability
Abstract: Advancements in robot capabilities are often achieved through integrating more hardware components. These hardware additions often lead to systems with high power consumption, fragility, and difficulties in control and maintenance. However, is this approach the only path to enhancing robot functionality? In this talk, I introduce the PuzzleBots, a modular multi-robot system with passive [...]
Design Principles for Robotics Systems that Support Human-Human Collaborative Learning
Abstract: Robots possess unique affordances granted by combining software and hardware. Most existing research focuses on the impact of these affordances on human-robot collaboration, but the theory of how robots can facilitate human-human collaboration is underdeveloped. Such theory would be beneficial in education. An educational device can afford collaboration in both assembly and use. This [...]
Leveraging Parallelism to Accelerate Quadratic Program Solvers for MPC
Abstract: Many problems in robotics can be formulated as quadratic programs (QPs). In particular, model-predictive control problems often involve repeatedly solving QPs at very high rates (up to kilohertz). However, while other areas of robotics like machine learning have achieved high performance by taking advantage of parallelism on modern computing hardware, state-of-the-art algorithms for solving [...]
Robust Incremental Distributed Collaborative Simultaneous Localization and Mapping
Abstract: Multi-robot teams show exceptional promise across applications like Search-and-Rescue, disaster-response, agriculture, forestry, and scientific exploration due to their ability to go where humans cannot, parallelize activity, operate robustly to failures, and expand capabilities beyond that of an individual robot. Collaborative Simultaneous Localization and Mapping (C-SLAM) is a fundamental capability for these multi-robot teams as [...]
Towards Equitable Representation in Text-to-Image Generation
Abstract: Accurate representation in media is known to improve the well-being of the people who consume it. There is a growing concern about the increasing use of generative AI in media as the generative image models trained on large web-crawled datasets such as LAION are known to produce images with harmful stereotypes and misrepresentations of various groups, [...]
3D Inference from Unposed Sparse View Images
Abstract: We propose UpFusion, a system that can perform novel view synthesis and infer 3D representations for generic objects given a sparse set of reference images without corresponding pose information. Current sparse-view 3D inference methods typically rely on camera poses to geometrically aggregate information from input views, but are not robust in-the-wild when such information [...]
Tightly Coupled LIDAR-Inertial Odometry
Abstract: In the age of self-driving, LIDAR and IMU represent two of the most ubiqui- tous sensors in use. Kalman Filtering and loosely coupled approaches dominate industry techniques, while current research trends towards a more tightly coupled formulation involving a joint optimization of IMU and LIDAR measurements. After two years of experience working with and [...]
A Unified Control Framework for Robust Aerial Manipulation
Abstract: Aerial robots are now widely employed in diverse applications, such as delivery, environmental monitoring, and especially aerial manipulation—the focus of this thesis. Aerial manipulation involves integrating robotic arms with drones to perform physical tasks remotely. This capability is particularly crucial for operations that are either too dangerous or inaccessible for humans, such as high-altitude [...]
In Pursuit of Open-World Mobile Manipulation
Abstract: Deploying robots in open-ended unstructured environments such as homes has been a long-standing research problem. However, robots are often studied only in closed-off lab settings, and prior mobile manipulation work is restricted to pick-move-place, which is arguably just the tip of the iceberg in this area. In this thesis, we introduce the Open-World Mobile [...]
Carnegie Mellon University
Geometric Heuristics Enhance POCUS AI for Pneumothorax
Abstract: The interpretation of Point-of-care ultrasound (POCUS) images poses a challenge due to the scarcity of high-quality labelled data for training AI models in the medical domain. To address this limitation, novel methodologies were developed to train POCUS AI models using limited data, integrating geometric heuristics derived from expert clinicians. Focused on diagnosing pneumothorax, heuristics [...]
Optimal Control and Robot Learning on Agile Safety-Critical Systems
Abstract: We present a pipeline of optimal control methods for learning an optimal control policy and locally accurate dynamics models for agile and safety-critical robots using autonomous racing as an application example. We introduce Spline-Opt, a fast offline/online optimization and planning method that can produce a reasonably good initial optimal trajectory given very little dynamics [...]
Vision Model Diagnosis and Improvement Via Large Pretrained Models
Abstract: As AI becomes increasingly pervasive in real-world applications, the deployment of machine learning models in real-world applications has underscored critical challenges in model robustness, fairness and performance. Despite significant advances, existing models often exhibit biases, fail to generalize across diverse data distributions, and struggle with unexpected input variations, leading to suboptimal or even discrimina- [...]
Beyond Robot Safety: Adaptability and Interactivity
Abstract: The deployment of autonomous robots in various areas, including transportation and human-robot collaboration, requires strong safety measures for effective interaction with the physical world. Traditional safe control algorithms work well in controlled settings but struggle to adapt to more interactive and unpredictable real-world scenarios. This thesis emphasizes the need to explore beyond traditional robot [...]
Indoor Localization and Mapping with 4D mmWave Imaging Radar
Abstract: State estimation is a crucial component for the successful implementation of robotic systems, relying on sensors such as cameras, LiDAR, and IMUs. However, in real-world scenarios, the performance of these sensors is degraded by challenging environments, e.g. adverse weather conditions and low-light scenarios. The emerging 4D imaging radar technology is capable of providing robust perception in adverse conditions. [...]
PIE-FRIDA: Personalized Interactive Emotion-Guided Collaborative Human-Robot Art Creation
Abstract: The introduction of generative AI has brought about many improvements in the artistic world. It allows many individuals to create artwork via simple descriptive text prompts. This has, in particular, created an avenue for non-artistic individuals to express their thoughts through generated art. Our work focuses on how emotion can be added as an [...]
Carnegie Mellon University
Spectral Mapping using Simple Sensors
Abstract: Spectral mapping holds significant importance in many exploration endeavors as it facilitates a deeper comprehension of material composition within a surveyed area. While imaging spectrometers excel in recording reflectance spectra into spectral maps, their large physical footprint, substantial power requirements, and operational intricacies render them unsuitable for integration into small rovers or resource-constrained missions. [...]
Simulated Encounters of the Third Kind: Scenario-Based Approach to Designing Guide Robots
Abstract: Navigating through unfamiliar environments is a challenging task. For people who are blind or have low vision (BLV), navigation can be particularly daunting. Guide robots are a type of service robot that can assist BLV people with navigation tasks. A significant amount of research related to guide robots has focused on technical contributions, while a [...]
Composing Generative and Discriminative Models for Better Generalization
Abstract: Computer Vision is Correspondence, correspondence, correspondence! Inspite of the singular definition of computer vision, we still have two broad categories of approaches in the literature. Generative Models, like Stable Diffusion, learn a correspondence between image and text modality, while learning a mapping from text to image. Discriminative Models, like CLIP, on the other hand [...]
Lower Bounds for Moving Target Traveling Salesman Motion Planning with Obstacles
Abstract: We study the problem of finding a trajectory for an agent to intercept a number of moving targets while avoiding obstacles. Applications include resupplying naval ships at sea and recharging aerial vehicles with a ground vehicle. We model the problem as an extension of the traveling salesman problem, which we refer to as the [...]
Towards Pragmatic Time Series Intelligence
Abstract: The widespread adoption of time series machine learning (ML) models faces multiple challenges involving data, modeling and evaluation. Data. Modern ML models depend on copious amounts of cohesive and reliably annotated data for training and evaluation. However, labeled data is not always available and reliable, and can also be dispersed across different locations. We [...]
Probabilistic 3D Multi-Object Cooperative Tracking for Autonomous Driving via Differentiable Multi-Sensor Kalman Filter
Abstract: Current state-of-the-art autonomous driving vehicles mainly rely on each individual sensor system to perform perception tasks. Such a framework's reliability could be limited by occlusion or sensor failure. To address this issue, more recent research proposes using vehicle-to-vehicle (V2V) communication to share perception information with others. However, most relevant works focus only on cooperative [...]
Super Odometry: Selective Fusion Towards All-degraded Environments
Abstract: Robust odometry is at the core of robotics and autonomous systems operating navigation, exploration, and locomotion in complex environments for a broad spectrum of applications. While great progress has been made, the robustness of the odometry system still remains a grand challenge. This talk introduces Super Odometry, an approach that leverages selective fusion to [...]
Improved Surface Estimation for use in Virtual Fixtures during Retinal Surgery
Abstract: Retinal surgery procedures require surgeons to manipulate very delicate tissues with little room for error. During epiretinal membrane surgery, to reduce chances of recurrence, surgeons may have to remove the 10 µm thick internal limiting membrane from the retinal surface. An experimental procedure to treat retinal vein occlusion is retinal vein cannulation. During this [...]
Causal Robot Learning for Manipulation
Abstract: Two decades into the third age of AI, the rise of deep learning has yielded two seemingly disparate realities. In one, massive accomplishments have been achieved in deep reinforcement learning, protein folding, and large language models. Yet, in the other, the promises of deep learning to empower robots that operate robustly in real-world environments [...]
Learning to Manipulate Using Diverse Datasets
Abstract: Autonomous agents can play games (like Chess, Go, and even Starcraft), they can help make complex scientific predictions (e.g., protein folding), and they can even write entire computer programs, with just a bit of prompting. However, even the most basic physical manipulation skills, like unlocking and opening a door, still remain literally out-of-reach. The [...]
3D Perception In-The-Wild
Abstract: State estimation is a fundamental component of embodied perception. Contemporary autonomous vehicle (AV) benchmarks have advanced techniques for training 3D detectors, particularly on large-scale data. Notably, although prior work has nearly solved 3D object detection for a few common classes (e.g., pedestrian and car), detecting many rare classes in-the-tail (e.g., debris and stroller) remains [...]
Learning on the Move: Integrating Action and Perception for Mobile Manipulation
Abstract: While there has been remarkable progress recently in the fields of manipulation and locomotion, mobile manipulation remains a long-standing challenge. Compared to locomotion or static manipulation, a mobile system must make a diverse range of long-horizon tasks feasible in unstructured and dynamic environments. While the applications are broad and interesting, there are a plethora [...]
Differentiable Convex Modeling for Robotic Planning and Control
Abstract: Robotic simulation, planning, estimation, and control, have all been built on top of numerical optimization. In this same time, modern convex optimization has matured into a robust technology delivering globally optimal solutions in polynomial time. With advances in differentiable optimization and custom solvers capable of producing smooth derivatives, convex modeling has become fast, reliable, [...]
Plan to Learn: Active Robot Learning by Planning
Abstract: Robots need a diverse repertoire of capable motor skills to succeed in the open world. Such a skillset cannot be learned or designed purely on human initiative. In this thesis, we advocate for an active continual learning approach that enables robots to take charge of their own learning. The goal of an autonomously learning [...]
Continual Personalization of Human Actions with Prompt Tuning
Abstract: In interactive computing devices (VR/XR headsets), users interact with the virtual world using hand gestures and body actions. Typically, models deployed in such XR devices are static and limited to their default set of action classes. The goal of our research is to provide users and developers with the capability to personalize their experience by [...]
Policy Decomposition
Abstract: Optimal Control is a popular formulation for designing controllers for dynamic robotic systems. Under the formulation, the desired long-term behavior of the system is encoded via a cost function and the policy, i.e. a mapping from the state of the system to control commands, to achieve the desired behavior are obtained by solving an [...]
Analysis by Synthesis for Modern Computer Vision
Abstract: Image denoising, depth completion, scene flow, and dynamic 3D reconstruction are all examples of recovery problems: the estimation of multidimensional signals from corrupted or partial measurements. This thesis examines these problems from the classic analysis-by-synthesis perspective, where a signal model is used to propose hypotheses, which are then compared to observations. This paradigm has [...]
Reinforcement Learning with Spatial Reasoning for Dexterous Robotic Manipulation
Abstract: Robotic manipulation in unstructured environments requires adaptability and the ability to handle a wide variety of objects and tasks. This thesis presents novel approaches for learning robotic manipulation skills using reinforcement learning (RL) with spatially-grounded action spaces, addressing the challenges of high-dimensional, continuous action spaces and alleviating the need for extensive training data. Our [...]
Leveraging Vision, Force Sensing, and Language Feedback for Deformable Object Manipulation
Deformable object manipulation represents a significant challenge in robotics due to its complex dynamics, lack of low-dimensional state representations, and severe self-occlusions. This challenge is particularly critical in assistive tasks, where safe and effective manipulation of various deformable materials can significantly improve the quality of life for individuals with disabilities and address the growing needs [...]
CBGT-Net: A Neuromimetic Architecture for Robust Classification of Streaming Data
Abstract: This research introduces CBGT-Net, a neural network model inspired by the cortico-basal ganglia-thalamic (CBGT) circuits in mammalian brains, which are crucial for critical thinking and decision-making. Unlike traditional neural network models that generate an output for each input or after a fixed sequence of inputs, CBGT-Net learns to produce an output once sufficient evidence [...]
Information-Based Adaptive Allocation of Heterogeneous Multi-Agent Teams for Search and Coverage
Abstract: Information-based search and coverage are important in planetary exploration and disaster response applications. Efficient information acquisition can help with increasing geological understanding or situational awareness. Heterogeneous robots, each with different sensing and motion modalities, can be coordinated to optimize search and coverage in a target region. Information maps, which estimate the importance of visiting [...]
Enhancing Robot Perception and Interaction Through Structured Domain Knowledge
Abstract: Despite the advancements in deep learning driven by increased computational power and large datasets, significant challenges remain. These include difficulty in handling novel entities, limited mechanisms for human experts to update knowledge, and lack of interpretability, all of which are crucial for human-centric applications like assistive robotics. To address these issues, we propose leveraging [...]
Dynamic Multi-Objective Trajectory Planning for Mobile Robots
Abstract: Robotic explorers play a crucial role in acquiring data from areas that are difficult or impossible for humans to reach. Whether for planetary exploration, search and rescue missions, agriculture, or other scientific exploration tasks, these robots can utilize pre-existing knowledge of the terrain to navigate effectively. In search- and coverage-oriented scenarios, robots must consider [...]
Motion planning for manipulation under pose uncertainty using contacts
Abstract: Numerous manipulation tasks, such as plug insertion and pipe assembly, demand an extremely high level of precision in pose estimation. Even minor errors, on the order of 2mm, can lead to task failure. While robots often rely on vision for object detection and localization, achieving consistent, high-precision localization using visual methods is not always [...]
Robust Off-road Wheel Odometry with Slip Estimation
Abstract: Wheel odometry is not often used in state estimation for off-road vehicles due to frequent wheel slippage, varying wheel radii, and the 3D motion of the vehicle not fitting with the 2D nature of integrated wheel odometry. This paper proposes a novel 3D preintegration of wheel encoder measurements on manifold. Our method additionally estimates [...]
Optimal Modular Robot Design for Mobile Manipulation in Agriculture
Abstract: Although agriculture is a highly mechanized industry, numerous sectors like horticulture and floriculture heavily depend on manual labor because they require safe handling of plants and produce that can only be left to humans. However, many research and commercial robots have succeeded in several challenging dexterous manipulation tasks like harvesting, pruning, and plant health [...]
Aligning Robot Task and Interaction Policies to Human Values
Abstract: The value alignment problem considers how robots can learn to behave in accordance with human values. Today, robot learning paradigms enable humans to provide data (e.g., preference labels or demonstrations), which the robot uses to update its behavior (e.g., reward model or policy) to be closer to the human’s values. However, the current paradigm [...]
Accelerating Robot Task Learning with Large Pretrained Models and Internet Data
Abstract: Large pre-trained models and internet data sources are key to general and efficient robot task learning. However, learning contact-rich behaviors, semantic task constraints, and robust task planning from internet data sources remains an open challenge. This proposal seeks to make progress towards a general robot task learning system leveraging pre-trained models and internet data. [...]
A Modularized Approach to Vision-based Tactile Sensor Design Using Physics-based Rendering
Abstract: Touch is an essential sensing modality for making autonomous robots more dexterous and allowing them to work collaboratively with humans. In particular, the advent of vision-based tactile sensors has resulted in efforts to design them for different robotic manipulation tasks. However, this design task remains a challenging problem. This is for two reasons: first, [...]
Towards Universal Place Recognition
Title: Towards Universal Place Recognition Abstract: Place Recognition is essential for achieving robust robot localization. However, current state-of-art systems remain environment/domain-specific and fragile. By leveraging insights from vision foundation models, we present AnyLoc, a universal VPR solution that performs across diverse environments without retraining or fine-tuning, significantly outperforming supervised baselines. We further introduce MultiLoc, and enable [...]
Enhancing Model Performance and Interpretability with Causal Inference as a Feature Selection Algorithm
Abstract: Causal inference focuses on uncovering cause-effect relationships from data, diverging from conventional machine learning which primarily relies on correlation analysis. By identifying these causal relationships, causal inference improves feature selection for predictive models, leading to predictions that are more accurate, interpretable, and robust. This approach proves especially effective with interventional data, such as randomized [...]
GNSS-denied Ground Vehicle Localization for Off-road Environments with Bird’s-eye-view Synthesis
Abstract: Global localization is essential for the smooth navigation of autonomous vehicles. To obtain accurate vehicle states, on-board localization systems typically rely on Global Navigation Satellite System (GNSS) modules for consistent and reliable global positioning. However, GNSS signals can be obstructed by natural or artificial barriers, leading to temporary system failures and degraded state estimation. On the [...]
Scaling up Robot Skill Learning with Generative Simulation
Abstract: Generalist robots need to learn a wide variety of skills to perform diverse tasks across multiple environments. Current robot training pipelines rely on humans to either provide kinesthetic demonstrations or program simulation environments with manually-designed reward functions for reinforcement learning. Such human involvement is an important bottleneck towards scaling up robot learning across diverse [...]
Simulation as a Tool for Conspicuity Measurement
Abstract: The use of unmanned aerial vehicles (UAVs) for time critical tasks is becoming increasingly popular. Operators are expected to use information from these swarms to make real-time and informed decisions. Consequently, detecting and recognizing targets from video is extremely pivotal to the success of these systems. At greater altitudes or with more vehicles, this [...]
VP4D: View Planning for 3D and 4D Scene Understanding
Abstract: View planning plays a critical role by gathering views that optimize scene reconstruction. Such reconstruction has played an important part in virtual production and computer animation, where a 3D map of the film set and motion capture of actors lead to an immersive experience. Current methods use uncertainty estimation in neural rendering of view [...]
Unlocking Generalization for Robotics via Modularity and Scale
Abstract: How can we build generalist robot systems? Looking at fields such as vision and language, the common theme has been large scale end-to-end learning with massive, curated datasets. In robotics, on the other hand, scale alone may not be enough due to the significant multimodality of robotics tasks, lack of easily accessible data and [...]
Automating Annotation Pipelines by leveraging Multi-Modal Data
Abstract: The era of vision-language models (VLMs) trained on large web-scale datasets challenges conventional formulations of “open-world" perception. In this work, we revisit the task of few-shot object detection (FSOD) in the context of recent foundational VLMs. First, we point out that zero-shot VLMs such as GroundingDINO significantly outperform state-of-the-art few-shot detectors (48 vs. 33 AP) [...]
Leveraging Affordances for Accelerating Online RL
Abstract: The inability to explore environments efficiently makes online RL sample-inefficient. Most existing works tackle this problem in a setting devoid of prior information. However, additional affordances may often be cheaply available at the time of training. These affordances include small quantities of demo data, simulators that can reset to arbitrary states and domain specific [...]
Dynamic Route Guidance in Vehicle Networks by Simulating Future Traffic Patterns
Abstract: Roadway congestion leads to wasted time and money and environmental damage. One possible solution is adding more roadway capacity, but this can be impractical especially in urban environments and still may not make up for a poorly-calibrated traffic signal schedule. As such, it is becoming increasingly important to use existing road networks more efficiently. [...]
Safe, Robust and Adaptive Model Learning for Agile Robots: Autonomous Racing
Abstract: In recent years there has been a rapid development in agile robots capable of operating at their limits in dynamic environments. Autonomous racing and recent developments in it also spurred by competitions such as the Indy Autonomous Challenge, A2RL, and F1Tenth have shown how modern autonomous control algorithms are capable of operating racecars at [...]
Improving Lego Assembly with Vibro-Tactile Feedback
Abstract: Robotic manipulation is an important area of research to improve the level of efficiency and autonomy in manufacturing processes. Due to the high precision and repeatability of industrial robot arms, robotic manufacturing tasks are dominated by simple pick, place, and peg insertion actions performed in a highly structured environment. Lego blocks are an excellent [...]
DeltaWalker: A Soft, Linearly Actuated Delta Quadruped Robot
Abstract: Quadruped robots offer a versatile solution for navigating complex terrain, making them valuable for applications such as industrial automation or search and rescue. Although quadrupeds are more complex than bipeds, they are easier to balance and control and require fewer joints to actuate compared to hexapods. Traditional quadruped designs, however, often feature complex leg [...]
Propagative Distance Optimization for Constrained Inverse Kinematics
Abstract: This work investigates a constrained inverse kinematic (IK) problem that seeks a feasible configuration of an articulated robot under various constraints such as joint limits and obstacle collision avoidance. Due to the high-dimensionality and complex constraints, this problem is often solved numerically via iterative local optimization. Classic local optimization methods take joint angles as [...]
Advancing Legged Robot Agility: from Video Imitation to GPU Acceleration
Abstract: Achieving human and animal-level agility has been a long-standing goal in robotics research. Recent advancements in numerical optimization and machine learning have pushed legged systems to greater capabilities than ever before, enabling black flips, parkour, and manipulation of heavy objects. Despite these exciting developments, this thesis identifies two key limitations of current legged robot [...]
Model Predictive Control on Resource-Constrained Robots
Abstract: Model predictive control (MPC) is a powerful tool for controlling highly dynamic robotic systems subject to complex constraints. However, it is computationally expensive and often requires a large memory footprint. Larger robotic systems are capable of carrying and powering sophisticated computational hardware onboard. On the other hand, smaller robots typically have faster dynamics that [...]
Enhancing Bipedal Locomotion With Reaction Wheels
Abstract: Legged robot hardware has become more accessible in the last ten years. However, there is still a dearth of low-cost hardware platforms that are open-source and easy to build. With recent developments in accessible manufacturing methods, such as 3D printing, it has become possible to design and manufacture parts without relying on precision machining. [...]
Building Micron: The Next Handheld Manipulator for Microsurgery
Abstract: Robotic assistance is used today in a variety of surgeries as a means of precise, dexterous, and minimally-invasive manipulation. However, practical use in microsurgical environments such as vitreoretinal surgery remains a challenge for the most common mechanically-grounded robotic platforms. Microsurgery requires micron-level accuracy and the ability to manipulate with interaction forces in millinewtons. Vitreoretinal [...]
Towards Estimation, Modeling, and Control of Mixed Material Flows on Variable-Speed Conveyor Belt Systems with Applications in Recycling
Abstract: Whether it is in sorting defects from grain in an agricultural setting, ore from tailings in a mine, or letters in a postal system, the sorting of bulk material has long been a crucial aspect of human industry. Today, in the face of dwindling natural resource deposits and accelerating climate change, a particularly important [...]
Expressive Attentional Communication Learning using Graph Neural Networks
Abstract: Multi-agent reinforcement learning presents unique hurdles such as the non-stationary problem beyond single-agent reinforcement learning that makes learning effective decentralized cooperative policies using an agent's local state extremely challenging. Effective communication to share information and coordinate is vital for agents to work together and solve cooperative tasks, as the ubiquitous evidence of communication in [...]
Estimating Object Importance and Modeling Driver’s Situational Awareness for Intelligent Driving
Abstract: The ability to identify important objects in a complex and dynamic driving environment can help assistive driving systems alert drivers. These assistance systems also require a model of the drivers' situational awareness (SA) (what aspects of the scene they are already aware of) to avoid unnecessary alerts. This thesis builds towards such intelligent driving [...]
Learning for Perception and Strategy: Adaptive Omnidirectional Stereo Vision and Tactical Reinforcement Learning
Abstract: Multi-view stereo omnidirectional distance estimation usually needs to build a cost volume with many hypothetical distance candidates. The cost volume building process is often computationally heavy considering the limited resources a mobile robot has. We propose a new geometry-informed way of distance candidates selection method which enables the use of a very small number [...]
Online-Adaptive Self-Supervised Learning with Visual Foundation Models for Autonomous Off-Road Driving
Abstract: Autonomous robot navigation in off-road environments currently presents a number of challenges. The lack of structure makes it difficult to handcraft geometry-based heuristics that are robust to the diverse set of scenarios the robot might encounter. Many of the learned methods that work well in urban scenarios require massive amounts of hand-labeled data, but [...]
Multimodal Representations for Adaptable Robot Policies in Human-Inhabited Spaces
Abstract: Human beings sense and express themselves through multiple modalities. To capture multimodal ways of human communication, I want to build adaptable robot policies that infer task pragmatics from video and language prompts, reason about sounds and other sensors, take actions, and learn mannerisms of interacting with people and objects. Existing solutions for robot policies [...]
Interleaving Discrete Search and Continuous Optimization for Kinodynamic Motion Planning
Abstract: Motion planning for dynamically complex robotic tasks requires explicit reasoning within constraints on velocity, acceleration, force/torque, and kinematics such as avoiding obstacles. To meet these constraints, planning algorithms must simultaneously make high-level discrete decisions and low-level continuous decisions. For example, pushing a heavy object involves making discrete decisions about contact locations and continuous decisions [...]