This article proposes a novel interaction force estimation scheme capable of estimating the interaction forces for industrial robots with high precision in the absence of force sensors. Specifically, we first establish the dynamic model of a class of industrial robots via model identification methods. Furthermore, a high-order finite-time observer (HOFTO) is established to estimate the interaction forces, where HOFTO serves as an interaction force observer and is designed to accurately estimate external time-varying interaction forces. In addition, we strictly analyze the stability of our approach in two different cases, endowing HOFTO with reliable estimation abilities in broader applications. Subsequently, we integrate the estimated force information into the applications of force control frameworks (e.g., collision detection and impedance control). Several evaluations on a real 6-axes robot demonstrate the effectiveness of the proposed approach.

Nowadays, the realization of obstacle avoidance for robot manipulators are generally based on offline path planning, which may be insufficient for real-time dynamic obstacle avoidance scenarios. In order to achieve safe and smooth avoidance of dynamic obstacles, a low-latency motion planning algorithm, which takes into account the dynamic planning is of practical significance. Toward this end, this article proposes a cascaded nonlinear model predictive control (MPC) assigned with a two-stage optimization problem. Specially, the high-level MPC combines artificial potential field as a motion planner to generate foresight smooth trajectories. The low-level MPC acts as a trajectory tracker and a safety protector, taking along hard constraints to avoid collisions and singularities. In addition, a super-twisting observer is deployed to enhance the motion estimation accuracy of dynamic obstacles. Experimental results show that the proposed approach is beneficial to achieve safe and smooth dynamic obstacle avoidance in real-world scenarios.

In vision-based bearing defect detection, random shapes and multiscale defects can result in inaccurate feature extraction, and complex backgrounds can interfere with defect localization. Therefore, a novel bearing defect detection network (BDDN) is proposed in this article, aiming to improve the accuracy of bearing defect detection, which consists of deformable ResNet50 network (DRN) and double attention feature pyramid network (DA-FPN). In the DRN, it is found that using deformable convolution for stages 1–3 of ResNet50 is able to extract features that are closer to the actual defect regions through rich experiments. The DA-FPN is employed to reduce the interference of complex backgrounds for defect detection, which consists of a feature-enhanced attention module (FEAM) and a feature-communication attention module (FCAM). The FEAM obtains more refined high-level semantic information in channel feature extraction, while the FCAM efficiently fuses defect features across different levels. Experimental results demonstrate that the BDDN effectively detects bearing cover surface defects, achieving a mean average precision of 87.9%. In addition, it outperforms several defect detection models on the public Northeastern University dataset (NEU-DET) in terms of accuracy.

This study presents an output feedback‐based sliding mode control approach for motion control systems with mismatched/matched disturbances. The key idea of this study is to reconstruct the unmeasured auxiliary states and disturbances synchronously online by constructing a higher‐order extended state observer (HOESO). It is shown that the resulting sliding mode controller based on HOESO can drive the system output to an ultimately bounded region, which can be regulated arbitrarily small by assigning a sufficiently large scaling gain. Different from the existing extended state observer methods, the newly proposed HOESO treats the known dynamics satisfying the Lipschitz condition and the unknown higher‐order disturbances separately. As such, it provides a feasible way for higher‐precision states/disturbance estimation. Thanks to the utilisation of HOESO, the upper bound of disturbances is not necessarily required as a priori and the switching gain is adaptively chosen based on the estimation error such that the chattering effects can be significantly reduced. Experiments for speed regulation of a brushless DC motor are conducted to verify the effectiveness of the proposed method.

Inertially stabilized platforms (ISPs), typically involving two or three axis gimbal, are extensively utilized to achieve accurate tracking of optical axis regardless of vehicle motion or vibration. Due to the complex nonlinearities induced by cross-couplings, parameter uncertainties, and external disturbances, high-performance control for ISPs is always a challenging task for practical applications, especially when the dynamic mass imbalance exists. In this brief, the situation that the mass distribution of an ISP is nonsymmetrical with respect to the mass center is emphatically analyzed, where the couplings are modeled and resolved into the known and unknown parts. To cope with the unmeasured states as well as the uncertainties composed of unknown couplings and remaining dynamics that are unmodeled, a higher order sliding mode observer (HOSMO) is developed for both the states and the lumped uncertainties estimation. By incorporating the estimated information, a robust continuous terminal sliding mode (TSM) control law is synthesized to admit the desired inertial angular rates tracking in finite time without offset. The validity of the presented control approach is demonstrated by experimental studies.

A flexible active safety motion (FASM) control approach is proposed for collision avoidance in robot manipulators. The key feature is the use of control barrier functions (CBFs) to design flexible CBF-guided safety criteria (CBFSC) with dynamically optimized decay rates, providing both flexibility and active safety in dynamic environments for robots. First, discrete-time CBFs are utilized to formulate the new flexible CBFSC with dynamic decay rates, which is then integrated into the model predictive control (MPC) framework. Notably, the decay rates of the CBFSC are incorporated as decision variables, allowing for dynamic adaptability during obstacle avoidance. In addition, a new cost function with an integrated penalty term is designed to dynamically adjust the safety margins. Finally, experiments in various scenarios using a Universal Robots 5 (UR5) manipulator validate the effectiveness of the proposed approach.