Yan Meng

In the field of multisensor fusion for simultaneous localization and mapping (SLAM), monocular cameras and IMUs are widely used to build simple and effective visual-inertial systems. However, limited research has explored the integration of motor-encoder devices to enhance SLAM performance. By incorporating such devices, it is possible to significantly improve active capability and field of view (FOV) with minimal additional cost and structural complexity. This article proposes a novel visual-inertial-encoder tightly coupled odometry (VIEO) based on a video detection and ranging (ViDAR) device. A ViDAR calibration method is introduced to ensure accurate initialization for VIEO. In addition, a platform motion decoupled active SLAM method based on deep reinforcement learning (DRL) is proposed. Experimental data demonstrate that the proposed ViDAR and the VIEO algorithm significantly increase cross-frame co-visibility relationships compared to its corresponding visual-inertial odometry (VIO) algorithm, improving state estimation accuracy. Additionally, the DRL-based active SLAM algorithm, with the ability to decouple from platform motion, can increase the diversity weight of the feature points and further enhance the VIEO algorithm's performance. The proposed methodology sheds fresh insights into both the updated platform design and decoupled approach of active SLAM systems in complex environments.

Gait planning is one of the most critical technologies for quadruped robots. However, far too little attention has been paid to the tight coupling mechanism of gait planning with terrain understanding and energy efficiency. To date, it is still challenging to plan optimal gait strategies that are highly adapted to terrain features with stable and efficient transitions. Accordingly, this paper proposes an adaptive gait control framework for quadruped robots that combines terrain recognition, Cost of Transport (CoT), and the Froude (Fr) number. More specifically, an optimal gait selection strategy for quadruped robots is designed based on different terrain texture features and the CoT characteristics of different gaits. To address the gait transition process induced thereby, an adaptive method for gait parameters based on the Fr number is further proposed, which can make the process more stable. Besides, model predictive control (MPC) and whole-body control (WBC) are employed as the motion controllers for the quadruped robot. Furthermore, simulation and experimental results indicate that the proposed method possesses superior terrain adaptability, energy efficiency, and motion stability during gait transitions, which is beneficial for the quadruped robots to maintain stable motion and reduce energy consumption when performing tasks in changeable terrains. Note to Practitioners —This paper is motivated by the problem of adaptive gait planning for quadruped robots that walks through different terrains. We propose a method that ensures optimal gait selection by robots facing diverse terrains and maintains the stability of gait transition. The proposed control framework, upon testing in a simulated environment, can be directly deployed on real-world robot without further adjustments and allows the robot to traverse various terrains with minimal sim-to-real issues. Hopefully, our proposed method can provide valuable guidance and support for facilitating the enhancement of capabilities in performing prolonged endurance tasks in unstructured environments for quadruped robots.

This paper proposes a novel tightly coupled monocular-inertial-pressure sensor fusion method for underwater localization of a biomimetic robotic manta. Based on the ORB-SLAM3 monocular visual-inertial odometry model, the depth measurement from a pressure sensor is incorporated and a novel approach is provided to associate low-frequency pressure measurements with high-frequency frames by utilizing only the relative depth between two adjacent keyframes as measurements. To address the challenge of scale estimation uncertainty in monocular odometry systems, a two-step monocular initialization strategy is proposed, involving an initial estimate based on visual-pressure measurements and a subsequent tightly coupled-inertial pressure depth residual construction, which results in a significantly improved scale estimate compared to conventional monocular inertial odometry systems. After successful initialization of the monocular system, a novel visual-inertial-pressure joint optimization method is proposed to enhance the localization and attitude estimation accuracy. Extensive experiments are carried out on both open-source datasets and real-world underwater datasets collected by a biomimetic robotic manta. The experimental results demonstrate the effectiveness of the proposed method in significantly improving both the position and attitude estimation of the biomimetic robotic manta. This research provides valuable insights for enhancing the underwater localization capability of biomimetic robotic systems.

Oral localization is widely used in oral and maxillofacial surgery. Most existing navigation systems adopt stereo optical devices to track fiducial markers for oral localization. However, they still suffer from line-of-sight occlusion and field-of-view limitation. This paper presents a robust oral localization system that utilizes multi-camera tracking of self-identifying markers to address the above issue. A customized oral clip with self-identifying markers on its curved surface is designed, which is firmly fixed within the patient’s oral cavity. Before surgery, an accurate method based on intraoral scanning is developed to register the patient’s oral cavity with the markers, of which the registration error is less than 0.1 mm. Secondly, a cylindrical marker is applied for calibrating the converging multi-camera system to avoid the blurry sight problem. During surgery, a multi-camera tracking method, integrating nonlinear optimization and an extended Kalman filter, is proposed to localize the oral cavity. We establish a simulated environment to test the accuracy and robustness of the tracking method. Additionally, we set up a real platform using a phantom oral cavity to validate the feasibility of each proposed method. Results reveal that our system yields a closer trajectory (0.37 mm) to the ground truth than that of and a stereo-camera tracking system (1.58 mm), and exhibits a certain robustness against camera occlusion.

Biological fish exhibit remarkable adaptability and exceptional swimming performance through their powerful and flexible bodies. Therefore, designing a continuum flexible body is significantly important for the development of a robotic fish. However, it is still challenging to replicate these functions of a biological body due to the limitations of actuation and material. In this paper, based on a tensegrity structure, we propose a bionic design scheme for a continuum robotic fish body with a property of stiffness variation. Its detailed structures and actuation principles are also presented. A mathematical model was established to analyze the bending characteristics of the tensegrity structure, which demonstrates the feasibility of mimicking the fish-like oscillation propulsion. Additionally, the stiffness variation mechanism is also exhibited experimentally to validate the effectiveness of the designed tensegrity fish body. Finally, a novel bionic robotic fish design scheme is proposed, integrating an electronic module-equipped fish head, a tensegrity body, and a flexible tail with a caudal fin. Subsequently, a prototype was developed. Extensive experiments were conducted to explore how control parameters and stiffness variation influence swimming velocity and turning performance. The obtained results reveal that the oscillation amplitude, frequency, and stiffness variation of the tensegrity robotic fish play crucial roles in swimming motions. With the stiffness variation, the developed tensegrity robotic fish achieves a maximum swimming velocity of 295 mm/s (0.84 body length per second, BL/s). Moreover, the bionic tensegrity robotic fish also performs a steering motion with a minimum turning radius of 230 mm (0.68 BL) and an angular velocity of 46.6°/s. The conducted studies will shed light on the novel design of a continuum robotic fish equipped with stiffness variation mechanisms.

In the field of quadruped robots, the most classic motion control algorithm is based on model prediction control (MPC). However, this method poses challenges as it necessitates the precise construction of the robot’s dynamics model, making it difficult to achieve agile movements similar to those of a biological dog. Due to these limitations, researchers are increasingly turning to model-free learning methods, which significantly reduce the difficulty of modeling and engineering debugging and simultaneously reduce real-time optimization computational burden. Inspired by the growth process of humans and animals, from learning to walk to fluent movements, this article proposes a hierarchical reinforcement learning framework for the motion controller to learn some higher-level tasks. First, some basic motion skills can be learned from motion data captured from a biological dog. Then, with these learned basic motion skills as a foundation, the quadruped robot can focus on learning higher-level tasks without starting from low-level kinematics, which saves redundant training time. By utilizing domain randomization techniques during the training process, the trained policy function can be directly transferred to a physical robot without modification, and the resulting controller can perform more biomimetic movements. By implementing the method proposed in this article, the agility and adaptability of the quadruped robot can be maximally utilized to achieve efficient operations in complex terrains.

Aquatic animals such as fish and cetaceans can actively modulate their body stiffness with muscle to achieve excellent swimming performance under different situations. However, it is still challenging for a robotic swimmer with bionic propulsion mode to dynamically adjust its body stiffness to improve the swimming speed due to the difficulties in designing an effective stiffness adjustment structure. In this paper, based on the special torque mode of a motor, we propose an active variable stiffness control method for a robotic dolphin to pursue better swimming speed. Different from a variable stiffness structure design, a torque control strategy for the caudal motor is employed to imitate the physical property of a torsion spring to act as the variable stiffness component. In addition, we also establish a dynamic model with the Lagrangian method to explore the variable stiffness mechanism. Extensive experiments have validated the dynamic model, and then the relationships between frequency and stiffness on swimming performance are presented. More importantly, through integrating the dynamic model and torque actuation mode-based variable stiffness mechanism, the online performance optimization scheme can be easily realized, providing valuable guidance in coordinating system parameters. Finally, experiments have demonstrated the stiffness adjustment capability of the caudal joint, validating the effectiveness of the proposed control method. The results also reveal that stiffness plays an essential role in swimming motion, and the active stiffness adjustment can significantly contribute to performance improvement in both speed and efficiency. Namely, with the adjustment of stiffness, the maximum speed of our robotic dolphin achieves up to 1.12 body length per second (BL/s) at 2.88 Hz increasing by 0.44 BL/s. Additionally, the efficiency is also improved by 37%. The conducted works will offer some new insights into the stiffness adjustment of robotic swimmers for better swimming performance.

The purpose of this article is to teach a robot assembly skills from demonstrations, and we attempt to train both the trajectory and the insertion force simultaneously. We encode human demonstrations data via motion primitives and, then, generate a reference trajectory and a prescribed force profile for a new assembly task using the combination of the motion primitives. We then propose an adaptive impedance controller to track the prescribed force with unknown environment stiffness, where the impedance parameters are estimated by the optimal solution of an equivalent linearization model. Our approach combines adaptive impedance control techniques with learning from demonstration on the same that makes it tractable and applicable to a real robot. Experiments on several typical noncylindrical parts illustrate the efficiency of the proposed method.

The inability to compare directly different nerve grafts has been a significant factor hindering the advance of nerve graft development. Due to the abundance of variables that exist in nerve graft construction and multiple assessment types, there has been limited success in comparing nerve graft effectiveness among experiments. Using mathematical techniques on nerve conduction velocity (NCV) autograft data, a normalization function was empirically derived that normalizes differences in gap lengths. Further analysis allowed for the development of the relative regeneration ratio (RRR). The RRR function allows researchers to directly compare nerve graft results based on the NCV data from their respective studies as long as the data was collected at the same post‐operation time. This function also allows for comparisons between grafts tested at different gap lengths. Initial testing of this RRR function provided confidence that the function is accurate for a continuum of gap lengths and different nerve graft types.