Organizational Unit:

School of Interactive Computing

Permanent Link

https://hdl.handle.net/1853/70783

Parent Organization

Organizational Unit

College of Computing

ArchiveSpace Name Record

https://finding-aids.library.gatech.edu/agents/corporate_entities/1113

Full item page

Publication Search Results

Now showing 1 - 10 of 20

Robotics in the Era of Vision-Language Foundation Models

(Georgia Institute of Technology, 2023-11-29) Kira, Zsolt
Foundation Models for Robotics

(Georgia Institute of Technology, 2023-11-29) Garg, Animesh
Robotics Days for Industry 2023 Welcome and Overview

(Georgia Institute of Technology, 2023-11-29) Hutchinson, Seth
Navigation behavior design and representations for a people aware mobile robot system

(Georgia Institute of Technology, 2016-01-15) Cosgun, Akansel

There are millions of robots in operation around the world today, and almost all of them operate on factory floors in isolation from people. However, it is now becoming clear that robots can provide much more value assisting people in daily tasks in human environments. Perhaps the most fundamental capability for a mobile robot is navigating from one location to another. Advances in mapping and motion planning research in the past decades made indoor navigation a commodity for mobile robots. Yet, questions remain on how the robots should move around humans. This thesis advocates the use of semantic maps and spatial rules of engagement to enable non-expert users to effortlessly interact with and control a mobile robot. A core concept explored in this thesis is the Tour Scenario, where the task is to familiarize a mobile robot to a new environment after it is first shipped and unpacked in a home or office setting. During the tour, the robot follows the user and creates a semantic representation of the environment. The user labels objects, landmarks and locations by performing pointing gestures and using the robot's user interface. The spatial semantic information is meaningful to humans, as it allows providing commands to the robot such as ``bring me a cup from the kitchen table". While the robot is navigating towards the goal, it should not treat nearby humans as obstacles and should move in a socially acceptable manner. Three main navigation behaviors are studied in this work. The first behavior is the point-to-point navigation. The navigation planner presented in this thesis borrows ideas from human-human spatial interactions, and takes into account personal spaces as well as reactions of people who are in close proximity to the trajectory of the robot. The second navigation behavior is person following. After the description of a basic following behavior, a user study on person following for telepresence robots is presented. Additionally, situation awareness for person following is demonstrated, where the robot facilitates tasks by predicting the intent of the user and utilizing the semantic map. The third behavior is person guidance. A tour-guide robot is presented with a particular application for visually impaired users.
Robots learning actions and goals from everyday people

(Georgia Institute of Technology, 2015-11-16) Akgun, Baris

Robots are destined to move beyond the caged factory floors towards domains where they will be interacting closely with humans. They will encounter highly varied environments, scenarios and user demands. As a result, programming robots after deployment will be an important requirement. To address this challenge, the field of Learning from Demonstration (LfD) emerged with the vision of programming robots through demonstrations of the desired behavior instead of explicit programming. The field of LfD within robotics has been around for more than 30 years and is still an actively researched field. However, very little research is done on the implications of having a non-robotics expert as a teacher. This thesis aims to bridge this gap by developing learning from demonstration algorithms and interaction paradigms that allow non-expert people to teach robots new skills. The first step of the thesis was to evaluate how non-expert teachers provide demonstrations to robots. Keyframe demonstrations are introduced to the field of LfD to help people teach skills to robots and compared with the traditional trajectory demonstrations. The utility of keyframes are validated by a series of experiments with more than 80 participants. Based on the experiments, a hybrid of trajectory and keyframe demonstrations are proposed to take advantage of both and a method was developed to learn from trajectories, keyframes and hybrid demonstrations in a unified way. A key insight from these user experiments was that teachers are goal oriented. They concentrated on achieving the goal of the demonstrated skills rather than providing good quality demonstrations. Based on this observation, this thesis introduces a method that can learn actions and goals from the same set of demonstrations. The action models are used to execute the skill and goal models to monitor this execution. A user study with eight participants and two skills showed that successful goal models can be learned from non- expert teacher data even if the resulting action models are not as successful. Following these results, this thesis further develops a self-improvement algorithm that uses the goal monitoring output to improve the action models, without further user input. This approach is validated with an expert user and two skills. Finally, this thesis builds an interactive LfD system that incorporates both goal learning and self-improvement and evaluates it with 12 naive users and three skills. The results suggests that teacher feedback during experiments increases skill execution and monitoring success. Moreover, non-expert data can be used as a seed to self-improvement to fix unsuccessful action models.
TAR: Trajectory adaptation for recognition of robot tasks to improve teamwork

(Georgia Institute of Technology, 2015-11-10) Novitzky, Michael

One key to more effective cooperative interaction in a multi-robot team is the ability to understand the behavior and intent of other robots. Observed teammate action sequences can be learned to perform trajectory recognition which can be used to determine their current task. Previously, we have applied behavior histograms, hidden Markov models (HMMs), and conditional random fields (CRFs) to perform trajectory recognition as an approach to task monitoring in the absence of commu- nication. To demonstrate trajectory recognition of various autonomous vehicles, we used trajectory-based techniques for model generation and trajectory discrimination in experiments using actual data. In addition to recognition of trajectories, we in- troduced strategies, based on the honeybee’s waggle dance, in which cooperating autonomous teammates could leverage recognition during periods of communication loss. While the recognition methods were able to discriminate between the standard trajectories performed in a typical survey mission, there were inaccuracies and delays in identifying new trajectories after a transition had occurred. Inaccuracies in recog- nition lead to inefficiencies as cooperating teammates acted on incorrect data. We then introduce the Trajectory Adaptation for Recognition (TAR) framework which seeks to directly address difficulties in recognizing the trajectories of autonomous vehicles by modifying the trajectories they follow to perform them. Optimization techniques are used to modify the trajectories to increase the accuracy of recognition while also improving task objectives and maintaining vehicle dynamics. Experiments are performed which demonstrate that using trajectories optimized in this manner lead to improved recognition accuracy.
Physics-based reinforcement learning for autonomous manipulation

(Georgia Institute of Technology, 2015-08-21) Scholz, Jonathan

With recent research advances, the dream of bringing domestic robots into our everyday lives has become more plausible than ever. Domestic robotics has grown dramatically in the past decade, with applications ranging from house cleaning to food service to health care. To date, the majority of the planning and control machinery for these systems are carefully designed by human engineers. A large portion of this effort goes into selecting the appropriate models and control techniques for each application, and these skills take years to master. Relieving the burden on human experts is therefore a central challenge for bringing robot technology to the masses. This work addresses this challenge by introducing a physics engine as a model space for an autonomous robot, and defining procedures for enabling robots to decide when and how to learn these models. We also present an appropriate space of motor controllers for these models, and introduce ways to intelligently select when to use each controller based on the estimated model parameters. We integrate these components into a framework called Physics-Based Reinforcement Learning, which features a stochastic physics engine as the core model structure. Together these methods enable a robot to adapt to unfamiliar environments without human intervention. The central focus of this thesis is on fast online model learning for objects with under-specified dynamics. We develop our approach across a diverse range of domestic tasks, starting with a simple table-top manipulation task, followed by a mobile manipulation task involving a single utility cart, and finally an open-ended navigation task with multiple obstacles impeding robot progress. We also present simulation results illustrating the efficiency of our method compared to existing approaches in the learning literature.
Semantic mapping for service robots: building and using maps for mobile manipulators in semi-structured environments

(Georgia Institute of Technology, 2015-04-08) Trevor, Alexander J. B.

Although much progress has been made in the field of robotic mapping, many challenges remain including: efficient semantic segmentation using RGB-D sensors, map representations that include complex features (structures and objects), and interfaces for interactive annotation of maps. This thesis addresses how prior knowledge of semi-structured human environments can be leveraged to improve segmentation, mapping, and semantic annotation of maps. We present an organized connected component approach for segmenting RGB-D data into planes and clusters. These segments serve as input to our mapping approach that utilizes them as planar landmarks and object landmarks for Simultaneous Localization and Mapping (SLAM), providing necessary information for service robot tasks and improving data association and loop closure. These features are meaningful to humans, enabling annotation of mapped features to establish common ground and simplifying tasking. A modular, open-source software framework, the OmniMapper, is also presented that allows a number of different sensors and features to be combined to generate a combined map representation, and enabling easy addition of new feature types.
Time-optimal sampling-based motion planning for manipulators with acceleration limits

(Georgia Institute of Technology, 2015-04-07) Kunz, Tobias

Robot actuators have physical limitations in how fast they can change their velocity. The more accurately planning algorithms consider these limitations, the better the robot is able to perform. Sampling-based algorithms have been successful in geometric domains, which ignore actuator limitations. They are simple, parameter-free, probabilistically complete and fast. Even though some algorithms like RRTs were specifically designed for kinodynamic problems, which take actuator limitations into account, they are less efficient in these domains or are, as we show, not probabilistically complete. A common approach to this problem is to decompose it, first planning a geometric path and then time-parameterizing it such that actuator constraints are satisfied. We improve the reliability of the latter step. However, the decomposition approach can neither deal with non-zero start or goal velocities nor provides an optimal solution. We demonstrate that sampling-based algorithms can be extended to consider actuator limitations in the form of acceleration limits while retaining the same advantageous properties as in geometric domains. We present an asymptotically optimal planner by combining a steering method with the RRT* algorithm. In addition, we present hierarchical rejection sampling to improve the efficiency of informed kinodynamic planning in high-dimensional spaces.
Autonomous environment manipulation to facilitate task completion

(Georgia Institute of Technology, 2015-03-31) Levihn, Martin

A robot should be able to autonomously modify and utilize its environment to assist its task completion. While mobile manipulators and humanoid robots have both locomotion and manipulation capabilities, planning systems typically just consider one or the other. In traditional motion planning the planner attempts to find a collision free path from the robot's current configuration to some goal configuration. In general, this process entirely ignores the fact that the robot has manipulation capabilities. This is in contrast to how humans naturally act - utilizing their manipulation capabilities to modify the environment to assist locomotion. If necessary, humans do not hesitate to move objects, such as chairs, out of their way or even place an object, such as a board, on the ground to reach an otherwise unreachable goal. We argue that robots should demonstrate similar behavior. Robots should use their manipulation capabilities to move or even use environment objects. This thesis aims at bringing robots closer to such capabilities. There are two primary challenges in developing practical systems that allow a real robotic system to tightly couple its manipulation and locomotion capabilities: the inevitable inaccuracies in perception as well as actuation that occur on physical systems, and the exponential size of the search space. To address these challenges, this thesis first extends the previously introduced domain of Navigation Among Movable Obstacles (NAMO), which allows a robot to move obstacles out of its way. We extend the NAMO domain to handle the underlying issue of uncertainty. In fact, this thesis introduces the first NAMO framework that allows a real robotic systems to consider sensing and action uncertainties while reasoning about moving objects out of the way. However, the NAMO domain itself has the shortcoming that it only considers a robot's manipulation capabilities in the context of clearing a path. This thesis therefore also generalizes the NAMO domain itself to the Navigation Using Manipulable Obstacles (NUMO) domain. The NUMO domain enables a robot to more generally consider the coupling between manipulation and locomotion capabilities and supports reasoning about using objects in the environment. This thesis shows the relationship between the NAMO and NUMO domain, both in terms of complexity as well as solution approaches, and presents multiple realizations of the NUMO domain. The first NUMO realization enables a robot to use its manipulation capabilities to assist its locomotion by changing the geometry of the environment for scenarios in which obstructions can be overcome through the usage of a single object. The system led a real humanoid robot to autonomously build itself a bridge to cross a gap and a stair step to get on a platform. A second NUMO realization then introduces reasoning about force constraints using knowledge about the mechanical advantages of a lever and battering ram. The discussed system allows a robot to consider increasing its effective force though the use of objects, such as utilizing a rod as a lever. Finally this thesis extends the NUMO framework for geometric constraints to scenarios in which the robot is faced with a substantial lack of initial state information and only has access to onboard sensing. In summary, this thesis enables robots to autonomously modify their environment to achieve task completion in the presence of lack of support for mobility, the need to increase force capabilities and partial knowledge.