Tendon-driven continuum robots (TDCR) are widely used in various engineering disciplines due to their exceptional flexibility and dexterity. However, their complex structure often leads to significant manufacturing costs and lengthy prototyping cycles. To cope with this problem, we propose a fused-deposition-modeling-printable (FDM-printable) TDCR structure design using a serial S-shaped backbone, which enables planar bending motion with minimized plastic deformation. A kinematic model for the proposed TDCR structure based on the pseudo-rigid-body model (PRBM) approach is developed. Experimental results have revealed that the proposed kinematic model can effectively predict the bending motion under certain tendon forces. In addition, analyses of mechanical hysteresis and factors influencing bending stiffness are conducted. Finally, A three-finger gripper is fabricated to demonstrate a possible application of the proposed TDCR structure.

A physically feasible, reliable, and safe motion is essential for robot operation. A parameterization-based trajectory planning approach is proposed for an 8-DOF manipulator with multiple constraints. The inverse kinematic solution is obtained through an analytical method, and the trajectory is planned in joint space. As such, the trajectory planning of the 8-DOF manipulator is transformed into a parameterization-based trajectory optimization problem within its physical, obstacle and task constraints, and the optimization variables are significantly reduced. Then teaching-learning-based optimization (TLBO) algorithm is employed to search for the redundant parameters to generate an optimal trajectory. Simulation and physical experiment results demonstrate that this approach can effectively solve the trajectory planning problem of the manipulator. Moreover, the planned trajectory has no theoretical end-effector deviation for the task constraint. This approach can provide a reference for the motion planning of other redundant manipulators.

Inspired by Model Predictive Interaction Control (MPIC), this paper proposes differential models for estimating contact geometric parameters and normal-friction forces and formulates an optimal control problem with multiple constraints to allow robots to perform rigid-soft heterogeneous contact tasks. Within the MPIC, robot dynamics are linearized, and Extended Kalman Filters are used for the online estimation of geometry-aware parameters. Meanwhile, a geometry-aware Hertz contact model is introduced for the online estimation of contact forces. We then implement the force-position coordinate optimization by incorporating the contact parameters and interaction force constraints into a gradient-based optimization MPC. Experimental validations were designed for two contact modes: “single-point contact” and “continuous contact”, involving materials with four different Young’s moduli and tested in human arm “relaxation-contraction” task. Results indicate that our framework ensures consistent geometry-aware parameter estimation and maintains reliable force interaction to guarantee safety. Our method reduces the maximum impact force by 50% and decreases the average force error by 42%. The proposed framework has potential applications in medical and industrial tasks involving the manipulation of rigid, soft, and deformable objects.

Position and velocity estimation are the key technologies to improve the motion control ability of humanoid robots. Aiming at solving the positioning problem of humanoid robots, we have designed a legged odometry algorithm based on forward kinematics and the feed back of IMU. We modeled the forward kinematics of the leg of the humanoid robot and used Kalman filter to fuse the kinematics information with IMU data, resulting in an accurate estimate of the humanoid robot’s position and velocity. This odometry method can be applied to different humanoid robots, requiring only that the robot is equipped with joint encoders and an IMU. It can also be extended to other legged robots. The effectiveness of the legged odometry scheme was demonstrated through simulations and physical tests conducted with the Walker2 humanoid robot.

In recent years, humanoid robots have gained significant attention due to their potential to revolutionize various industries, from healthcare to manufacturing. A key factor driving this transformation is the advancement of visual perception systems, which are crucial for making humanoid robots more intelligent and autonomous. Despite the progress, the full potential of vision-based technologies in humanoid robots has yet to be fully realized. This review aims to provide a comprehensive overview of recent advancements in visual perception applied to humanoid robots, specifically focusing on applications in state estimation and environmental interaction. By summarizing key developments and analyzing the challenges and opportunities in these areas, this paper seeks to inspire future research that can unlock new capabilities for humanoid robots, enabling them to better navigate complex environments, perform intricate tasks, and interact seamlessly with humans.

Soft robots can exhibit better performance in specific tasks compared to conventional robots, particularly in healthcare related tasks. However, the field of soft robotics is still young, and designing them often involves mimicking natural organisms or relying heavily on human experts’ creativity. A formal automated design process is required. The use of neuroevolution-based algorithms to automatically design initial sketches of soft actuators that can enable the movement of future medical devices, such as drug-delivering catheters, is proposed. The actuator morphologies discovered by algorithms like Age-Fitness Pareto Optimisation, NeuroEvolution of Augmenting Topologies (NEAT), and Hypercube-based NEAT (HyperNEAT) were compared based on the maximum displacement reached and their robustness against various control methods. Analysing the results granted the insight that neuroevolution-based algorithms produce better-performing and more robust actuators under diverse control methods. Specifically, the best-performing morphologies were discovered by the NEAT algorithm.

Recent advancements in robotics have transformed industries such as manufacturing, logistics, surgery, and planetary exploration. A key challenge is developing efficient motion planning algorithms that allow robots to navigate complex environments while avoiding collisions and optimizing metrics like path length, sweep area, execution time, and energy consumption. Among the available algorithms, sampling-based methods have gained the most traction in both research and industry due to their ability to handle complex environments, explore free space, and offer probabilistic completeness along with other formal guarantees. Despite their widespread application, significant challenges still remain. To advance future planning algorithms, it is essential to review the current state-of-the-art solutions and their limitations. In this context, this work aims to shed light on these challenges and assess the development and applicability of sampling-based methods. Furthermore, we aim to provide an in-depth analysis of the design and evaluation of ten of the most popular planners across various scenarios. Our findings highlight the strides made in sampling-based methods while underscoring persistent challenges. This work offers an overview of the important ongoing research in robotic motion planning.

This paper presents a novel method for learning force-aware robot assembly skills, specifically targeting the peg insertion task on inclined hole. For the peg insertion task involving inclined holes, we employ one-dimensional convolutional networks (1DCNN) and gated recurrent units (GRU) to extract features from the time-series force information during the assembly process, thereby identifying different contact states between the peg and the hole. Subsequent to the identification of contact states, corresponding pose adjustments are executed, and overall smooth interaction is ensured through admittance control. The assembly process is dynamically adjusted using a state machine to fine-tune admittance control parameters and seamlessly switch the assembly state. Through the utilization of dual-arm clamping, we conduct key unlocking experiments on bases inclined at varying degrees. Our results demonstrate that the proposed method significantly improves the accuracy and success rate of state recognition compared to previous methods.

Humans excel at dexterous manipulation; however, achieving human-level dexterity remains a significant challenge for robots. Technological breakthroughs in the design of anthropomorphic robotic hands, as well as advancements in visual and tactile perception, have demonstrated significant advantages in addressing this issue. However, coping with the inevitable uncertainty caused by unstructured and dynamic environments in human-like dexterous manipulation tasks, especially for anthropomorphic five-fingered hands, remains an open problem. In this paper, we present a focused review of human-like dexterous manipulation for anthropomorphic five-fingered hands. We begin by defining human-like dexterity and outlining the tasks associated with human-like robot dexterous manipulation. Subsequently, we delve into anthropomorphism and anthropomorphic five-fingered hands, covering definitions, robotic design, and evaluation criteria. Furthermore, we review the learning methods for achieving human-like dexterity in anthropomorphic five-fingered hands, including imitation learning, reinforcement learning and their integration. Finally, we discuss the existing challenges and propose future research directions. This review aims to stimulate interest in scientific research and future applications.

The ability of bipedal humanoid robots to walk adaptively on varied terrain is a critical challenge for practical applications, drawing substantial attention from academic and industrial research communities in recent years. Traditional model-based locomotion control methods have high modeling complexity, especially in complex terrain environments, making locomotion stability difficult to ensure. Reinforcement learning offers an end-to-end solution for locomotion control in humanoid robots. This approach typically relies solely on proprioceptive sensing to generate control policies, often resulting in increased robot body collisions during practical applications. Excessive collisions can damage the biped robot hardware, and more critically, the absence of multimodal input, such as vision, limits the robot’s ability to perceive environmental context and adjust its gait trajectory promptly. This lack of multimodal perception also hampers stability and robustness during tasks. In this paper, visual information is added to the locomotion control problem of humanoid robot, and a three-stage multi-objective constraint policy distillation optimization algorithm is innovantly proposed. The expert policies of different terrains to meet the requirements of gait aesthetics are trained through reinforcement learning, and these expert policies are distilled into student through policy distillation. Experimental results demonstrate a significant reduction in collision rates when utilizing a control policy that integrates multimodal perception, especially in challenging terrains like stairs, thresholds, and mixed surfaces. This advancement supports the practical deployment of bipedal humanoid robots.

Neural networks have demonstrated exceptional performance across a range of applications. Yet, their training often demands substantial time and data resources, presenting a challenge for autonomous robots operating in real-world environments where real-time learning is difficult. To mitigate this constraint, we propose a novel human-in-the-loop framework that harnesses human expertise to mitigate the learning challenges of autonomous robots. Our approach centers on directly incorporating human knowledge and insights into the robot’s learning pipeline. The proposed framework incorporates a mechanism for autonomous learning from the environment via reinforcement learning, utilizing a pre-trained model that encapsulates human knowledge as its foundation. By integrating human-provided knowledge and evaluation, we aim to bridge the division between human intuition and machine learning capabilities. Through a series of collision avoidance experiments, we validated that incorporating human knowledge significantly improves both learning efficiency and generalization capabilities. This collaborative learning paradigm enables robots to utilize human common sense and domain-specific expertise, resulting in faster convergence and better performance in complex environments. This research contributes to the development of more efficient and adaptable autonomous robots and seeks to analyze how humans can effectively participate in robot learning and the effects of such participation, illuminating the intricate interplay between human cognition and artificial intelligence.