A5101849424- Formal Methods in Verification
- Autonomous Vehicle Technology and Safety
- Distributed systems and fault tolerance
- Robotic Path Planning Algorithms
- Software Reliability and Analysis Research
- Vehicle Dynamics and Control Systems
- Real-time simulation and control systems
- Software Testing and Debugging Techniques
- Robotics and Sensor-Based Localization
- Robotic Locomotion and Control
- Prosthetics and Rehabilitation Robotics
- AI-based Problem Solving and Planning
- Advanced Software Engineering Methodologies
- Balance, Gait, and Falls Prevention
- Fault Detection and Control Systems
- Petri Nets in System Modeling
- UAV Applications and Optimization
- Traffic Prediction and Management Techniques
- Human-Automation Interaction and Safety
- Robotics and Automated Systems
- Real-Time Systems Scheduling
- Modular Robots and Swarm Intelligence
- Security and Verification in Computing
- Machine Learning and Algorithms
- Underwater Vehicles and Communication Systems
Shenyang Jianzhu University
2024
Georgia Institute of Technology
2012-2022
Atlanta Technical College
2022
Baidu (China)
2020
Fordham University
2015
Texas A&M University
2011-2014
Beihang University
2012
Heterogeneous teams of robots, leveraging a balance between autonomy and human interaction, bring powerful capabilities to the problem exploring dangerous, unstructured subterranean environments. Here we describe solution developed by Team CSIRO Data61, consisting CSIRO, Emesent Georgia Tech, during DARPA Subterranean Challenge. These presented systems were fielded in Tunnel Circuit August 2019, Urban February 2020, our own Cave event, conducted September 2020. A unique capability team is...
This letter presents a free space trajectory optimization algorithm for autonomous driving, which decouples the collision-free generation problem into Dual-Loop Iterative Anchoring Path Smoothing (DL-IAPS) and Piecewise-Jerk Speed Optimization (PJSO) problem. The work leads to remarkable driving performance improvements including more robust precise collision avoidance, higher control feasibility, computation efficiency stricter comfort guarantee, compared with other existing algorithms....
The objectives of this article are: 1) to present a taxonomy for mixed-initiative human-robot interaction and 2) survey its state practice through the examination past research along each taxonomical dimension. paper starts with some definitions (MII) from perspective human-computer (HCI) introduce basic concepts MII. We then synthesize these robotic context human robot teams. A in is presented. goal inform design systems by identifying key elements systems. surveyed examined
While robots have long been proposed as a tool to reduce human personnel's exposure danger in subterranean environments, these environments also present significant challenges the development of robots. Fundamental this challenge is problem autonomous exploration. Frontier-based methods powerful and successful approach exploration, but complex 3D remain when online employment required. This paper presents new that addresses complexity operating by directly modelling boundary between observed...
Certain robot missions need to perform predictably in a physical environment that may have significant uncertainty. One approach is leverage automatic software verification techniques establish performance guarantee. The addition of an model and uncertainty both program environment, however, means the state space model-checking solution problem can be prohibitively large. An based on behavior-based controllers process-algebra framework avoids state-space combinatorics presented here. In this...
This paper presents an optimization-based collision avoidance trajectory generation method for autonomous driving in free-space environments, with enhanced robustness, comfort and efficiency. Starting from the hybrid framework, we introduce two warm start methods, temporal dual variable starts, to improve We also reformulates problem robustness name this new algorithm TDR-OBCA. With these changes, compared original optimization achieve a 96.67% failure rate decrease respect initial...
This paper presents a method to determine outputs associated with human walking data that can be used design controllers achieve human-like robotic walking. We consider collection of outputs, i.e., functions the kinematics computed from experimental data, satisfy criteria necessary for human-inspired bipedal robot control construction. These are described in form amendable controller through special class time based - termed canonical functions. An optimization problem is presented...
This paper describes the need and methods required to construct an integrated software verification mission specification system for use in robotic missions intended counter-weapons of mass destruction (c-WMD) operations, as part a 3-year effort Defense Threat Reduction Agency. The overall architecture is described. principal tool process algebra, PARS, based on port automata theory. PARS introduced, emphasizing its ability represent probabilistic programs uncertain dynamic environments,...
Certain robot missions need to perform predictably in a physical environment that may only be poorly characterized advance. We have previously developed an approach establishing performance guarantees for behavior-based controllers process-algebra framework. extend work here include random variables, and we show how our prior results can used generate Dynamic Bayesian Network the coupled system of program model. Verification is reduced filtering problem this network. Finally, present...
This work seeks virtual constraints, or outputs, that are intrinsic to human walking and utilises these outputs construct controllers which produce human-like bipedal robotic walking. Beginning with experimental data, sought, i.e., functions of the kinematics over time, provides a low-dimensional representation It will be shown that, for humans act like linear mass-spring-dampers; this yields time through canonical functions. Combining formulations leads human-inspired when utilised in an...
Deploying a robot as part of counter-weapons mass destruction mission demands that the robotic software operates with high assurance. A unique feature development is need to perform predictably in physical environment may only be poorly characterized advance. In this paper, we present an approach building assurance for missions carried out uncertain environments. The framework and verification algorithm, VIPARS, are described detail. Results presented including motion sensing uncertainty,...
The effective use of autonomous robot teams in highly-critical missions depends on being able to establish performance guarantees. However, establishing a guarantee for the behavior an operating uncertain environment with obstacles is challenging problem. This paper addresses challenges involved building software tool verifying multi-robot waypoint mission that includes geometry as well uncertainty motion. One contribution this approach problem a-priori specification environments program...
We have developed an approach that can be used by mission designers to determine whether or not a performance guarantee for their software, when carried out under the uncertain conditions of real-world environment, will hold within threshold probability. In this paper we demonstrate its utility verifying multirobot missions, in particular bounding overwatch mission.
To better understand human walking, three bipedal robotic models-starting with the compass gait biped and increasing in complexity to a 3D kneed biped-are studied controllers of human-inspired design; these are derived from experimental data measuring kinematics test subjects. The collected examined an attempt classify some most fundamental behaviors underlying walking; it is found that subset functions on humans can be represented as single class functions. control scheme uses feedback...
Knowledge is essential for an autonomous robot to act intelligently when tasked with a mission. With recent leaps of progress, the paradigm SLAM (Simultaneous Localization and Mapping) has emerged as ideal source spatial knowledge robots. However, despite advancements in both paradigms control, research integration these areas been lacking remained open investigation. This paper presents into behavior-based robotic system dynamically acquired memory, which can be used enable new behaviors...
Autonomous robots offer the potential to conduct Counter-Weapons of Mass Destruction (C-WMD) missions in an efficient and robust manner. However, leverage this potential, a mission designer needs be able determine how well robot system will operate noisy uncertain environments that C-WMD may require. We are developing software framework for verification performance guarantees based on MissionLab novel process algebra approach representing programs operating environments.
Robotics has been considered as one of the five key technology areas for defense against attacks with weapons mass destruction (WMD). However, due to impact nature WMD, failures counter-WMD (C-WMD) missions can have catastrophic consequences. To ensure robots' success in carrying out C-WMD missions, we developed a novel verification framework providing performance guarantees behavior-based and probabilistic robot algorithms complex real-world environments. This paper describes system...
This paper presents a free space trajectory optimization algorithm of autonomous driving vehicle, which decouples the collision-free planning problem into Dual-Loop Iterative Anchoring Path Smoothing (DL-IAPS) and Piece-wise Jerk Speed Optimization (PJSO). The work leads to remarkable performance improvements including more precise collision avoidance, higher control feasibility better comfort, as those are often hard realize in other existing path/speed decoupled methods. Our algorithm's...
This paper presents the design of a tune-free (human-out-of-the-loop parameter tuning) control framework, aiming at accelerating large scale autonomous driving system deployed on various vehicles and environments. The framework consists three machine-learning-based procedures, which jointly automate tuning for driving, including: learning-based dynamic modeling procedure, to enable control-in-the-loop simulation with highly accurate vehicle dynamics tuning; open-loop mapping solve...
Retrosynthesis analysis is pivotal yet challenging in drug discovery and organic chemistry. Despite the proliferation of computational tools over past decade, AI-based systems often fall short generalizing across diverse reaction types exploring alternative synthetic pathways. This paper presents BatGPT-Chem, a large language model with 15 billion parameters, tailored for enhanced retrosynthesis prediction. Integrating chemical tasks via unified framework natural SMILES notation, this...
The heating energy consumption in public buildings cold regions is notably significant, presenting substantial scope for savings and emission reductions. Flexible loads can actively participate controlling the operation of power grid, improving utilization economy system. This study introduces flexible into optimization systems, establishing mathematical models thermal electrical loads. A two-stage method proposed: first stage simulates starting stopping control conditions equipment at...