Title: Trust-Integrated Intelligent Control for Socially Aware Robot Navigation in Dynamic and Crowded Environments (PI)
Funding Agency: National Science and Technology Council (NSTC), Taiwan
Duration: Aug. 2025 ~ Jul. 2029
Description: The primary goal of this project is to develop a socially-aware and trust-integrated mobile robot system to address the challenges of intelligent navigation and control in dynamic and crowded environments. With the rapid growth of urbanization and smart technologies, the application of mobile robots in public space services and logistics has gained increasing attention. However, when robots operate in densely populated and structurally complex environments, achieving natural collaboration with humans while enhancing trust and safety remains a critical issue. This project leverages the Social Force Model (SFM), integrating multi-modal sensing technologies, reinforcement learning, imitation learning, and passivity theory to build an intelligent navigation and control system capable of modeling heterogeneous crowd behaviors and enabling natural human-robot interactions. The research focuses on heterogeneous mobile robots, including wheeled and bipedal humanoid robots, with an in-depth exploration of motion planning, behavior modeling, and safety navigation strategies in crowded environments. Key research components include constructing and applying a generalized Social Force Model, designing navigation strategies based on crowd dynamics, developing natural gait control technologies, and studying collaborative behaviors in multi-robot systems. Additionally, the project will employ deep learning and large-scale data analysis to create a crowd dynamics simulation platform, analyze heterogeneous crowd characteristics, and design behavior-adaptive control algorithms. To enhance public acceptance and trust in mobile robot systems, the project will introduce safety indicators such as predicted time-to-collision (PTTC) to improve the interactive experience between robots and crowds from a smart navigation and control perspective. Through simulations and experimental validation, the project aims to ensure the practical value and stability of the proposed systems. Emphasis will be placed on improving the harmony, social compatibility, and interaction within heterogeneous crowd-robot collective dynamic systems. The research outcomes are expected to significantly advance the development and application of mobile robots, accelerating their integration into everyday life and fostering trustworthy interactions. Beyond academic contributions, this project is anticipated to have a profound impact on industrial applications, providing robust theoretical and technical support for mobile robot systems and human-robot collaboration. Furthermore, the project's achievements will contribute to the development of smart cities and artificial intelligence technologies in Taiwan, enhancing the nation's competitiveness and influence in global technological innovation. During the project, active collaboration with leading international scholars in robotics, dynamic systems, human-robot interaction, and artificial intelligence will be pursued. By fostering deep international academic exchanges and inviting scholars to participate in domestic academic activities, the project aims to maximize its impact and dissemination, while elevating Taiwan's importance in the global advancement of mobile robotics.
Title: Intelligent AI-Driven Outdoor Agricultural Autonomous Robot Technology Development and Integration (PI)
Funding Agency: National Science and Technology Council (NSTC), Taiwan
Duration: Jun. 2025 ~ May 2026
Description: Agriculture in Taiwan is facing challenges such as labor shortages, an aging workforce, and climate change, particularly in fruit harvesting operations, which are highly labor-intensive and complex, limiting agricultural productivity. This project focuses on autonomous fruit harvesting in outdoor orchards by developing smart agricultural robotics technology. By integrating Vision-Language Models (VLMs), Artificial Intelligence (AI), Machine Learning (ML), intelligent autonomous navigation, and robotic arm technology, this project aims to enhance fruit recognition and harvesting efficiency, reduce labor dependency, and advance agricultural automation and intelligence. This project consists of three subprojects and a main project. Sub-Project 1 focuses on the development of smart grippers and robotic arm motion control, incorporating force feedback and slip detection technology to improve harvesting success rates and minimize fruit damage. Sub-Project 2 specializes in multimodal sensing and machine learning models, utilizing VLMs and deep learning techniques to enhance the robot's ability to recognize fruit maturity and optimize intelligent localization for the mobile platform, ensuring precise harvesting decisions. Sub-Project 3 is responsible for autonomous navigation and the mobile robotic arm platform, employing Model Predictive Path Integral (MPPI) and reinforcement learning techniques to develop robots capable of autonomous movement and obstacle avoidance in orchard environments, enabling them to adapt flexibly to various terrains and tree structures. The Main project integrates the technologies from all sub-projects, ensuring that the robot possesses environmental perception, decision-making, and precise operation capabilities. Additionally, by leveraging VLMs and Large Language Models (LLMs), the system enhances intelligent decision-making, allowing the robot to optimize harvesting actions and movement strategies based on real-time environmental conditions. The anticipated research outcomes of this project will break through existing limitations in agricultural robotics by utilizing AI-driven technologies and machine learning to create a highly intelligent, precise, and adaptable fruit harvesting robot, driving Taiwan's agricultural automation and digital transformation. In the future, the developed technologies can be further applied to pruning, pollination, and pest monitoring, expanding the application potential of smart agricultural robots in various domains and fostering the transition of Taiwan's agriculture toward a more efficient, sustainable, and intelligent future.
Title: AI-Driven Multimodal Socially Interactive Robots for Adaptive Behaviors and Cross-Cultural Interaction (PI)
Funding Agency: National Cheng Kung University (NCKU), Taiwan
Duration: Aug. 2025 ~ Jul. 2028
Description: This project focuses on developing AI-driven multimodal socially interactive robots with intelligent behavior adaptation and cross-cultural contextual awareness. By integrating generative AI, deep reinforcement learning, multimodal AI sensing, agentic AI, physical AI, and the social force model, the goal is to enable natural interaction and intelligent navigation of robots in real-world public environments. The research covers "socially-driven AI for robot motion interaction and navigation," emphasizing how robots can generate socially aware behavioral decisions and collision avoidance strategies based on crowd dynamics and social norms. Additionally, the project incorporates "emotionally-driven AI for language and cultural interaction," allowing robots to automatically adjust explanations, tone, and non-verbal cues for users with different language proficiencies and cultural backgrounds to enhance contextual understanding and trust building. A cross-cultural human interaction database and simulation platform will also be developed to support AI model training and validation. The project brings together expertise in AI, robot control, human-robot interaction, and language-cultural studies, with international leading experts joining the core NCKU research team. By uniting top domestic and international teams, this initiative aims to promote interdisciplinary AI innovation, enhance intelligent robot capabilities, and contribute meaningfully to human-robot integration in smart cities and aging societies.
Title: Designing and Implementing Generative AI Service Robots: Exploring Applications in Companionship and Home Services (co-PI)
Funding Agency: National Science and Technology Council (NSTC), Taiwan
Duration: Aug. 2024 ~ Jul. 2026
Description: At present, science and technology are advancing rapidly. The emergence of generative artificial intelligence technology has given many possibilities to the field of robotics. Because of the shortage of human resources in modern society, the number of people who need to be accompanied and served is gradually increasing. The need for robots with humanized characteristics and the ability to interact with people is also rising. Therefore, this project will build a companion robot integration system based on generative AI, so that the robot can naturally integrate into human life and provide humanized companionship experience. This research project is expected to complete the construction of a companion robot and its integrated system based on generative AI in three years. In the first year, we will focus on designing and building a 6-DOF robot manipulator, and establishing a multi-sensor system at the same time. It will also integrate dialogue management based on generative AI and task planning method based on large language models. The next step is completing the integration of generative AI visual recognition of current state and data augmentation. After integration, the system will be verified with the accurate execution of commands given by user to the robotic arm. In the second year, the chassis mechanism will be developed and integrated with the aforementioned robotic manipulator to complete the overall robot mechanism. This year will apply TAIDE for system development. An environment perception and mobility system will be established. Then the human-computer interaction test including dialogue, vision, and cognitive functions will be carried out. After that, the entire robot function will be validated. In the third year, we will optimize various functions of the robot system, including GAI based the human-computer interaction and cognitive decision-making system. As well as integration and implementation of optimized systems for following, companionship, and motion control, will conduct. After completing the final system integration, the validation task will extend to the scenario across indoor and outdoor fields. It ensures that the robot system can provide reliable and intelligent services in various environments and situations.
Title: Human-Robot Interaction Systems Using Haptics Feedback Technology within the Cyber-Physical World (PI)
Funding Agency: National Science and Technology Council (NSTC), Taiwan
Duration: Aug. 2023 ~ Jul. 2026
Description: Due to the influence of the Covid-19 epidemic in the past few years, all companies and institutes keeps figuring out novel ways to maintain functional operations invulnerable to the lockdown and isolation by the governments. These tremendous demands lead to significant developments of new technologies using communication networks and virtual reality techniques so that human-robot interaction over a long distance can be much more feasible. Along with the trends, the integration of haptic technology, computer vision, and cyber-physical systems has attracted attention from scientific and technological circles. Although the key techniques for providing vision and sound in a virtual environment are on-going research problem, compared with Haptics feedback generated through physical interaction, most of the commercial VR systems is inadequate to provide satisfactory tactile experiences in virtual worlds. Therefore, there have been significant studies focusing on the combination of Haptics technology with virtual reality, allowing users to feel both the vision/sound and touch/force in a virtual environment. The academic community that has invested a lot of manpower and resources in the development of control theory, system integration, and intelligent technology, can be expected to play an important and key role in this emerging development direction, especially in cyber-physical systems, robotic systems, human-robot interaction, and Haptics technology. To provide better interaction, imagination, and immersion in human-robot teleoperation within a virtual environment over cyberspace, the ultimate goal of this four-year project is to not only improve an intuitive way of remote interaction with robotic systems but also use Haptic technology and virtual environment to ensure a more realistic and immersive human-robot interaction. By the end of this project, it is anticipated that the emerging technology will be highly developed to be beneficial for both the necessary technologies on Metaverse and novel system integration of cyber-physical systems, virtual reality, control theory, and robotics.
Title: Distributed Resilient Control for Persistent Coverage and Connectivity Maintenance in Heterogeneous Multi-Robot Systems (PI)
Funding Agency: National Science and Technology Council (NSTC), Taiwan
Duration: Aug. 2024 ~ Jul. 2025
Description: With the rapid development of technology and the widespread adoption of artificial intelligence in recent years, the control systems of multi-robot system, integrated through a large number of individual robots, have become a crucial research direction in both domestic and international fields of automatic control and robotics. Leveraging their high integration, scalability, applicability, and adaptability, these agent-based collaborative systems enhance autonomy and intelligence in complex systems, focusing on decision-making patterns, spatial awareness, and task design. Exploring advanced technologies and applications for multi-robot system is the central focus of this research project. For multi-robot system, the project initially concentrates on the development of distributed consensus control strategies and fixed-time convergence observers. This ensures that multiple robots are not restricted by geographical limitations, thereby optimizing performance dependent on mission-critical state information. Regarding communication architecture, the project delves into the local connectivity, global connectivity, and topological connectivity of multi-robot systems. It also addresses uncertainties in network communication, such as deception attacks, replay attacks, and disruption attacks, by proposing a highly resilient control architecture and strategies. On the application front, the project focuses on persistent coverage control. It aims to deepen the understanding of system convergence and the performance of optimized coverage. Additionally, the project integrates ground and aerial heterogeneous multi-robot systems into persistent coverage research, enhancing collaboration within the decentralized control framework. This integration significantly boosts the development potential and foundation of multi-robot systems in both theoretical and practical applications. In summary, with the continuous expansion of application scopes and increasing system complexity, multiple robots with agent-based collaborative systems have become a key research focus. This project aims to pioneer advanced technologies and applications for multi-robot systems by addressing challenges in distributed control, communication resilience, and persistent coverage, ultimately enhancing the development potential and foundation of multi-robot system in various theoretical and practical aspects.
Title: Advanced Resilient and Intelligent Control for Heterogeneous Cyber-Physical Mobile Robot Systems (PI) - Pilot Directions for MOST Grant for the Columbus Program
Funding Agency: Ministry of Science and Technology (MOST), Taiwan
Duration: Feb. 2019 ~ Jul. 2024
Description: With the advanced development of technology and the rapid flow of communication network, control system, and technology has an increasingly inseparable connection to social operation. Along with wide applications and system complexities, parameter uncertainties, environmental disturbance, communication attacks, or sensor/actuator malfunctions will lead to difficulty in the design and implementation of complex control systems. Recently, control engineers have raised and adopted the idea of resilience in the development and study of control systems. Resilience, which is known to human society but less known to engineering, is a property of a system, measuring the ability of a system to recover from partial damage on its own resource. In control engineering, the purpose of resilience is to develop advanced control frameworks and algorithms to make the entire system have the ability to endure, cope with, and recover from the aforementioned loss of efficacy. Moreover, a resilient control system must have the ability to ensure a minimum acceptable performance when the entire system is under the influence of various malfunctions from the environment, plant, and communication. It is expected that a resilient control system will recover from its original function when the unexpected damages are removed. For the development of next-generation robotic systems, robots must be endowed with cognitive sensitivity, maneuverable mobility, long-distance communication, and adaptive ability to new environments. The concept of resilient and intelligent control with cyber-physical framework is highly expected to be the cornerstone of future robotic systems. In order to enhance robotic systems with the ability to adapt to, deal with, and recover from various system, environment, and communication malfunctions, the ultimate goal of this five-year project is to develop and design advanced resilient and intelligent control frameworks and algorithms for multiple mobile robots under the cyber-physical system. The heterogeneity of grounded mobile robots, aerial mobile robots, mobile manipulators, and human-robot interaction will all be considered and studied in this project. By the end of this project, it is anticipated that the emerging technology will be highly developed for the control and robotic society in Taiwan, the outcomes of this Columbus project will become a significant foundation worldwide in these research topics, and more importantly, this project is expected to pilot and lead the relevant societies in Taiwan to enter the new world of cyber-physical resilient and intelligent mobile robot control systems.
Title: System Design of Integrated Magnetic Gear Motor Drive for Electric Vehicles (co-PI)
Funding Agency: Ministry of Science and Technology (MOST), Taiwan
Duration: Oct. 2015 ~ Sep. 2019
Description: Designating to hybrid electric vehicle (HEV) and EV, this project proposes a high performance and efficiency power train system which is capable to achieve energy-saving and environment-friendly goals. To reach the desired purpose, the design of an integral magnetic gear motor (IMGM) is developed and its applications in HEV and EV are investigated, which the efficiency of the power train and the operated range of the motor could be increased; meanwhile, the gearbox is compactly integrated and designed in the motor, thus the weight and volume of the power train system are effectively reduced. Thus, the cruising endurance of the EV and HEV is enhanced and raised higher than traditional ones. Further, the high performance driver and controller are adopted to maintain the motor’s operation in high efficiency even with different loading. Moreover, the new designs of the accelerometer and cam mechanism are employed to implement a system platform that can detect and reduce vibration generated by vehicle motors. According to the operating characteristics of gear motors, the design and analysis of the control strategies are examined for the novel developed EV to possess high stability, controllability, safety, and energy saving. In addition, combining the concept of reliability and optimality, a multi-parameter coupling simulation analysis system is established for fast and accurate design and evaluation of IMGMs. Finally, a big data database is set up to analyze the data of vehicle motor development and the equipment, monitoring the trend of system aging or damaging, and make recommendations for speed parameters. This project is divided into six sub-projects, covering electromagnetic analysis and design, drive control, piezoelectric material and vibration suppression analysis, control strategies of EV and HEV, multi-parameter coupling simulation, network teaching and patent platform, which expecting to deep down the industrial fundamental technology in vehicle motor development.
Title: Cooperation and Human-Robot Interaction for Multiple Mobile Manipulators in Cyber-Physical Control Systems (PI)
Funding Agency: Ministry of Science and Technology (MOST), Taiwan
Duration: Aug. 2017 ~ Jul. 2020 (terminated at Apr. 30, 2019 due to the requirement of MOST Columbus Program)
Description: With advanced development of technology and rapid flow of information, the variety of products has increased with shorter life-cycle. To confront with such challenges, only customized products can be able to satisfy most of the customers’ demand. In contract to traditional production line, the production process and systems should be more dexterous and flexible so that the change of products will be more adaptable to maintain higher efficiency and effectiveness. The idea of Industry 4.0 is to upgrade the industrial production process nowadays to the next stage with better machine utilization and faster throughput times. In order to enhance and accomplish the development of intelligent manufacturing for the foreseeable future of Industry 4.0, in order to improve manufacturing systems in Taiwan with advanced technology and configurations, the ultimate goal of this three-year project is to develop and design an advanced control framework and algorithm for multiple mobile manipulators under cyber-physical systems for Industry 4.0 and intelligent manufacturing. The control problems, e.g. coordination control, cooperation control, human-robot interaction, and virtual reality, will all be studied for multiple mobile manipulators in cyber-physical system to achieve various missions in industry. In the first year, the novel control framework and algorithms for multiple mobile manipulators in cyber-physical systems will be developed by taking into account the communication unreliability and limited bandwidth. Distributed cooperation control for collaboratively transporting a common object will be studied in the second year by utilizing leader-follower and abstract task function control schemes. Human-robot interaction for multiple mobile manipulators will be investigated in the third with the techniques of bilateral teleoperation and virtual reality. This project will take the advantages of cyber-physical system and mobile manipulators to accomplish the requirement and vision of robotic system in flexibility, scalability, interactivity, and expansibility for the future of Industry 4.0 and intelligent manufacturing.
Title: Cloud-Based Coordination and Cooperation for Mobile Robot Networks in Indoor Environment (PI)
Funding Agency: Ministry of Science and Technology (MOST), Taiwan
Duration: Aug. 2016 ~ Jul. 2019 (terminated at Apr. 30, 2019 due to the requirement of MOST Columbus Program)
Description: The ultimate goal of this three- year project is to develop and design an intelligent control system for mobile robot networks that takes into account the integration of cloud computation, cloud database, Internet of Things, and human-robot interaction. In contrast to independent robotic systems, cloud architecture can endue robots with the sensing capability to get aware of the environment and objects surrounding a robot, with the thinking capability to process the sensory results, and with the reactive ability to execute the action. In this project, sensing information acquired from wireless sensors, wearable devices, and mobile robots will be integrated into a function called Sensing Density Function, SDF. This function that represents the importance and priority of the environment will be utilized in control law for the mobile robot networks. In the first year, the control issues of coverage control and task allocation will be developed under cloud and IoT infrastructure. Patrolling control and robot deployment within an indoor environment will be studied in the second year. By integrating the research results in the first two years, the problem of path planning and collision avoidance for mobile robot networks will be investigated in the third year. Furthermore, the proposed control framework and algorithms will be validated via experiments using a group of mobile robot in indoor environment. This project will take the advantages of high-speed processing from cloud computers and ubiquitous information from IoT system to establish high-intelligent and high-performance multiple mobile robot system.
Title: Robot Control in Cyber-Physical Systems (PI)
Funding Agency: Ministry of Science and Technology (MOST), Taiwan
Duration: Aug. 2014 ~ Jul. 2017
Description: Cyber-physical system (CPS) is a novel system framework that endues physical systems with powerful computational abilities and abundant knowledge via the infrastructure of ubiquitous networks around the world. By connecting to the internet worldwide, physical systems become a portion of the communication nodes and have the ability to access other computer nodes in the cyberspace. The development of CPS has received immense interest from numerous research groups and could bring various potential applications. For the study of robotic systems, the integration with information from communication network, computational abilities from super computer, and interaction from the human beings would lead to unforeseen applications that are not achievable by using traditional robots. The field of robotic systems is growing tremendously. Many companies and research institutes have started to invest large amounts of funds in developing a variety of robots. Even though robots with the fruitful modern technologies are much more powerful than before, limited information, insufficient mobility, and high power consumption retrain the potential of future development. Therefore, this project will investigate the fundamental control algorithms and develop system framework to integrate robotic systems into cyber-physical system. In the design of the cyber-physical control framework, the properties of heterogeneity, compositionality, and scalability will be taken into account in the presence of network unreliability. Subsequently, the proposed cyber-physical robotic system will be applied to interactive human-robot teleoperation system and distributed visual-servoing robot control system.
Title: Study on Control Strategy and Failure Mode Analysis for Plug-in Hybrid Electric Vehicles (PI)
Funding Agency: Yen Tjing Ling Industrial Research Institute, Taiwan
Duration: Jan. 2015 ~ Dec. 2015
Description: Plug-in hybrid electric vehicle (PHEV) is a kind of hybrid electric vehicles (HEV) that has a large capacity battery which can provide short distance for commuters. PHEV has two kinds of power sources, internal combustion engine (ICE) and electric motor (EM). In order to have a good fuel economy, a control strategy for spilt the power between the engine and the motor is important. In this project, we develop a novel distribution strategy to regulate output power from ICE and EM. The main objective is to decide whether ICE have to be started to provide driving torque. A PD-like control algorithm is first proposed to guarantee that state of charge (SOC) for battery decreases to the minimum acceptable value. Although the total power within the battery can be totally utilized during a driving cycle, the performance of emission and fuel economy is worse than other methods. Therefore, in this project, we will design a model predictive control (MPC)-based control strategy for a parallel PHEV to not only extract power from battery but also improve emission and fuel economy. Simulation results by using MATLAB/Simulink will be performed to validate the proposed strategy.
Title: Control on Bilateral Human-Swarm Interaction over Communication Networks: Theory and Experiments (PI)
Funding Agency: National Science Council (NSC), Taiwan
Duration: Aug. 2013 ~ Jul. 2014
Description: A bilateral teleoperation system, composed of a master and a slave robot exchanging signals over a long distance communication, has been demonstrated to be a useful tool in implementing tasks in remote or hazardous environment. The development of teleoperation system could potentially contribute to a variety of application, such as remote medical operations and undersea explorations. However, most of the previous research on teleoperation system can only be utilized for the framework of one-master and one-slave sophisticated robots to cope with well-designed missions. Contrarily, the study of multi-robot system has been shown a powerful method to achieve complicated tasks without using sophisticated robotic systems. However, the control framework of multi-robot system considers the scenario in the absence of human operator, and can only be applied to accomplish pre-defined tasks with a lack of flexibility. Since enduing a group of swarm robots with the intelligence of human operators is useful in various applications and has attracted significant attention from controls and robotics communities, in this proposal the PI will develop a system framework and investigate the control problem of bilateral human-swarm interaction which could possess the advantages of teleoperation systems and swarm robot systems. In order to achieve teleoperation between kinematic dissimilarly local manipulator and remote swarm robots, the idea of controlling task-space function will be addressed in this project. The human operator is able to teleoperate a group of swarm by manipulating the corresponding task-space function of the local robotic manipulator. Since the teleoperation system is achieved in the task space and the behavior of entire swarm robots performs like a redundant manipulator, the additional degree-of-freedom of the swarm robot can be utilized to accomplish secondary tasks autonomously. Thus, the group of swam robot can regulate their configurations to adapt to the remote environment in order to provide a better teleoperation performance. In addition, time delays resulting from transmitting signals over a long distance communication network can pose significantly impediments to the stabilization problem and potentially degrade the performance of the closed-loop control system. Therefore, the stability of the proposed system with both constant and time-varying delays will be considered in this project. Simulations and experiments will be addressed to validate the proposed system and control algorithms.
Title: On Stability and Tracking Performance for Control of Robotic Systems over Communication Network (PI)
Funding Agency: National Science Council (NSC), Taiwan
Duration: Oct. 2012 ~ Sep. 2013
Description: With the infrastructure of ubiquitous networks around the world, the study of robotic systems under communication networks has drawn considerable attention from various communities. This novel system framework, compared to tradition wired connections between robots, controllers, actuators, and human operators, could bring significant advantages to the existent robotic systems, such as increased flexibility, ease of maintenance, and decreased costs. However, signals transmitted over long distances are generally subjected to time delays, packet loss, and data reordering. These communication unreliabilities can pose significantly impediments to the stabilization problem and potentially degrade the performance of the closed-loop control system. In order to guarantee system performance and stability, the objective of this project is to develop system frameworks for control of robotic systems over communication network under the influence of packet loss and gravitational force. Based on the passivity property of the robotic dynamics and appropriately defined controllers, the scattering representation, which was originally developed for bilateral teleoperation system, will be utilized to ensure stability and performance of the networked robotic systems. The influence of gravity in the robotic system under input/output time delays will be considered first in this project. Under the assumption that the gravitational model is known, a new control algorithm will be proposed in conjunction with delayed position feedback to guarantee both stability and tracking performance. In the second part of this project, a packet-based scattering transformation will be developed to ensure the passivity of communication channels under packet loss. In order to guarantee system performance under gravitational force over packet-switching communication, a passive position modulation will be proposed in the end of this project. The proposed control algorithms will be validated via simulations and experiments on robotic manipulators.