This paper set out to investigate the usefulness of solving collision avoidance problems with the help of deep reinforcement learning in an unknown environment, especially in compact spaces, such as a narrow corridor. This research aims to determine whether a deep reinforcement learning-based collision avoidance method is superior to the traditional methods, such as potential field-based methods and dynamic window approach. Besides, the proposed obstacle avoidance method was developed as one of the capabilities to enable each robot in a novel robotic system, namely the Self-reconfigurable and Transformable Omni-Directional Robotic Modules (STORM), to navigate intelligently and safely in an unknown environment. A well-conceived hardware and software architecture with features that enable further expansion and parallel development designed for the ongoing STORM projects is also presented in this work. A virtual STORM module with skid-steer kinematics was simulated in Gazebo to reduce the gap between the simulations and the real-world implementations. Moreover, comparisons among multiple training runs of the neural networks with different parameters related to balance the exploitation and exploration during the training process, as well as tests and experiments conducted in both simulation and real-world, are presented in detail. Directions for future research are also provided in the paper.
27 February 2019 in Journal of Mechanisms and Robotics
This paper describes the use of an active disturbance rejection controller (ADRC) to estimate and compensate for the effect of slip in an online manner to improve the path tracking performance of autonomous ground vehicles (AGVs). AGVs with skid-steer locomotion mode are extensively used for robotic applications in the fields of agriculture, transportation, construction, warehouse maintenance, and mining. Majority of these applications such as performing reconnaissance and rescue operations in rough terrain or autonomous package delivery in urban scenarios, require the system to follow a path predetermined by a high-level planner or based on a predefined task. In the absence of effective slip estimation and compensation, the AGVs, especially tracked vehicles, can fail to follow the path as given out by the high-level planner. The proposed ADRC architecture uses a generic mathematical model that can account for the scaling and shift in the states of the system due to the effects of slip through augmented parameters. An extended Kalman filter (EKF) observer is used to estimate the varying slip parameters online. The estimated parameters are then used to compensate for the effects of slip at each iteration by modifying the control actions given by a low-level path tracking controller. The proposed approach is validated through experiments over flat and uneven terrain conditions including asphalt, vinyl flooring, artificial turf, grass, and gravel using a tracked skid-steer mobile robot. A detailed discussion on the results and directions for future research is also presented.
1 February 2019 in Journal of Intelligent & Robotic Systems
One of the most impactful and exciting applications of robotic technology, especially autonomous and semi-autonomous systems, is in the field of search and rescue. Robots present an opportunity to go where rescuers cannot, keep responders out of danger, work indefatigably, and augment the capabilities of the humans who put their lives at risk while helping others. This paper examines the use of robotic systems in human rescue applications, with an emphasis on performing search, extraction, evacuation, and medical field treatment procedures. The work begins with a review of the various robotic systems designed to perform one or more of the above operations. The relative merits of each system are discussed along with their shortcomings. The paper also addresses the use of robotic competitions as a means of benchmarking field robotic systems. Based on the review of state of the art systems, a novel concept (Semi- Autonomous Victim Extraction Robot) designed to address the shortcomings of existing systems is described in the conclusion, along with detailed discussion on how it improves upon state of the art systems. The future research thrusts to be explored before realizing a fully integrated robotic rescue system are also detailed.
10 September 2019 in Journal of Mechanisms and Robotics
This paper describes the design and control of a novel hand exoskeleton. A subcategory of upper extremity exoskeletons, hand exoskeletons have promising applications in healthcare services, industrial workplaces, virtual reality, and military. Although much progress has been made in this field, most of the existing systems are position controlled and face several design challenges, including achieving minimal size and weight, difficulty enforcing natural grasping motions, exerting sufficient grip strength, ensuring the safety of the users hand, and maintaining overall user friendliness. To address these issues, this paper proposes a novel, slim, lightweight linkage mechanism design for a hand exoskeleton with a force control paradigm enabled via a compact series elastic actuator. A detailed design overview of the proposed mechanism is provided, along with kinematic and static analyses. To validate the overall proposed hand exoskeleton system, a fully integrated prototype is developed and tested in a series of experimental trials.
29 January 2018 in Journal of Mechanisms and Robotics
This paper presents the design and integration of a two-digit robotic exoskeleton glove mechanism. The proposed glove is designed to assist the user with grasping motions, such as the pincer grasp, while maintaining a natural coupling relationship among the finger and thumb joints, resembling that of a normal human hand. The design employs single degree-of-freedom (DOF) linkage mechanisms to achieve active flexion and extension of the index finger and thumb. This greatly reduces the overall weight and size of the system making it ideal for prolonged usage. The paper describes the design, mathematical modeling of the proposed system, detailed electromechanical design, and software architecture of the integrated prototype. The prototype is capable of recording information about the index finger and thumb movements, interaction forces exerted by the finger/thumb on the exoskeleton, and can provide feedback through vibration. In addition, the glove can serve as a standalone device for rehabilitation purposes, such as assisting in achieving tip or pulp pinch. The paper concludes with an experimental validation of the proposed design by comparing the motion produced using the exoskeleton glove on a wooden mannequin with that of a natural human hand.
13 July 2021 in Journal of Mechanisms and Robotics
This paper presents a self-reconfigurable modular robot with an integrated two degrees-of-freedom (DOF) active docking mechanism. Active docking in modular robotic systems has received a lot of interest recently as it allows small versatile robotic systems to coalesce and achieve the structural benefits of large systems. This feature enables reconfigurable modular robotic systems to bridge the gap between small agile systems and larger robotic systems. The proposed self-reconfigurable mobile robot design exhibits dual mobility using a tracked drive for longitudinal locomotion and a wheeled drive for lateral locomotion. The 2-DOF docking interface allows for efficient docking while tolerating misalignments. To aid autonomous docking, visual marker-based tracking is used to detect and re-position the source robot relative to the target robot. The tracked features are then used in Image-Based Visual Servoing to bring the robots close enough for the docking procedure. The hybrid-tracking algorithm allows eliminating external pixelated noise in the image plane resulting in higher tracking accuracy along with faster frame update on a low-cost onboard computational device. This paper presents the overall mechanical design and the integration details of the modular robotic module with the docking mechanism. An overview of the autonomous tracking and docking algorithm is presented along with a proof-of-concept real-world demonstration of autonomous docking and self-reconfigurability. Experimental results to validate the robustness of the proposed tracking method, as well as the reliability of the autonomous docking procedure, are also presented.