Robotics and Autonomous Systems最新文献

The present paper aims at introducing a mathematical model of a spherical robot expressed in the language of Lie-group theory. Since the main component of motion is rotational, the space $\mathrm{SO}\left(3\right)3$ of three-dimensional rotations plays a prominent role in its formulation. Because of friction to the ground, rotation of the external shell results in translational motion. Rolling without slipping implies a constraint on the tangential velocity of the robot at the contact point to the ground which makes it a non-holonomic dynamical system. The mathematical model is obtained upon writing a Lagrangian function that describes the mechanical system and by the Hamilton minimal-action principle modified through d’Alembert virtual work principle to account for non-conservative control actions as well as frictional reactions. The result of the modeling appears as a series of non-holonomic Euler–Poincaré equations of dynamics plus a series of auxiliary equations of reconstruction and advection type. A short discussion on the numerical simulation of such mathematical model complements the main analytic-mechanic development.

3D object detection is a key element for the perception of autonomous vehicles. LiDAR sensors are commonly used to perceive the surrounding area, producing a sparse representation of the scene in the form of a point cloud. The current trend is to use deep learning neural network architectures that predict 3D bounding boxes. The vast majority of architectures process the LiDAR point cloud directly but, due to computation and memory constraints, at some point they compress the input to a 2D Bird’s Eye View (BEV) representation. In this work, we propose a novel 2D neural network architecture, namely the Feature Aware Re-weighting Network, for feature extraction in BEV using local context via an attention mechanism, to improve the 3D detection performance of LiDAR-based detectors. Extensive experiments on five state-of-the-art detectors and three benchmarking datasets, namely KITTI, Waymo and nuScenes, demonstrate the effectiveness of the proposed method in terms of both detection performance and minimal added computational burden. We release our code at https://github.com/grgzam/FAR.

Recent years have experienced a notable surge in unmanned aerial vehicles (UAV) research, prompting exploration into innovative concepts. This paper introduces a compact UAV harnessing the Coandă effect, an underexplored phenomenon in fluid mechanics. Featuring a single lift motor and two types of flaps, this UAV offers exceptional maneuverability, presenting significant challenges compared to conventional multi-rotor UAVs. To address these challenges, we explore the theoretical study, mechatronic design, and manufacturing complexities of the Coandă UAV. Emphasizing the distinctiveness of our work, we assess a Fuzzy Logic Controller (FLC) for UAV stabilization, marking the first application of such techniques to a Coandă-effect UAV, in contrast to the Proportional-Integral-Derivative (PID) control employed by other researchers. This innovative application of Fuzzy logic, particularly the Sugeno model, proves advantageous, offering faster and more robust control in uncertain or noisy environments. The proposed FLC strategy is systematically compared against a classical PID control approach, formulated based on the Mamdani and Sugeno models, optimized and manually tuned using a genetic algorithm. Our results showcase a significantly improved settling time of 0.417 s with the FLC strategy, surpassing the PID control approach by 35.23%. To substantiate our findings, we present comprehensive experimentation conducted at both software and hardware levels using Matlab® and Simulink for a microcontroller-based UAV. This groundbreaking fusion of novel design and advanced control techniques not only addresses the unique challenges posed by the Coandă UAV's aerodynamics but also contributes significantly to the field of UAV research.

This paper presents the virtual design of a four-arm Delta parallel robot for potential applications in the food industry, specifically for the automated preparation of fast foods. Kinematic and dynamic models were developed based on the morphology of this type of robot. Cutting-edge and previously unexplored control strategies for these types of manipulators are then designed and implemented based on Fuzzy PD and Fuzzy PID configurations. Several performance indicators, such as the Integral Square Error (ISE), Integral Time Absolute Error (ITAE) and Integral Time Square Error (ITSE), were used to conduct a performance comparison of the control techniques, considering type-1 and interval type-2 fuzzy sets. In all the analyzed scenarios, the fuzzy controllers correspond to the Takagi–Sugeno–Kang model using linear functions of the inputs in the outputs. Among the main contributions of this work is the development of a detailed dynamic model of the robot in Simscape, incorporating realistic aspects that are often overlooked during analytical modeling. To ensure more accurate results, the rejection of external disturbances is thoroughly analyzed in simulations, considering elements such as time delays and control signal saturations. The results demonstrate the veracity of the proposed design for a four-arm Delta robot, both in models and in the Simscape/Simulink implementation. In the trajectory tracking task and external disturbance rejection, the superiority of the Fuzzy PID controller with interval type-2 fuzzy sets over fuzzy controllers with PD structure and over type-1 fuzzy sets is evidenced.

Simulating a soft robot model is challenging because of large deformations and lots of degrees of freedom. In this study, a novel fast simulation framework for soft robots is created using reduced order extended position-based dynamics technique. This framework is helpful for configuration design and motion verification. The soft robot model is built using strain constraints, and the reduced order matrix is created by linear modes and modal derivatives. It simplifies the creation of models and improves the efficiency of the reduced order matrix construction. To verify the feasibility of the present method, soft robots with various actuation and material combinations are simulated. The results also agree well with both the physical experiments and the finite element analysis.

In this paper, a new approach is proposed for the smooth path planning of Ackermann mobile robots based on an improved ant colony algorithm and B-spline curves. Firstly, by incorporating path length constraints and path smoothing constraints into the objective function, the smooth path planning problem for Ackermann mobile robots is transformed into a multi-objective optimization problem. Secondly, to address the limitations of the traditional ant colony algorithm, an improved ant colony algorithm based on the turning angle constraint (IACO-TAC) is proposed. IACO-TAC incorporates the distance factor and steering angle penalty factor in the heuristic function to reduce the path search's blindness. Moreover, the pheromone update method is improved, consisting of local pheromone update and global pheromone update, which uses a reward penalty mechanism to improve the convergence speed of the algorithm and increases the pheromone concentration of the global optimal path, respectively. Thirdly, an improved B-spline curve smoothing algorithm that considers the minimum turning radius constraint is proposed to generate a path that satisfies the kinematic constraints of the Ackermann mobile robot. Finally, the proposed method is evaluated by conducting gradient comparison experiments and ant colony algorithm comparison experiments on maps of different sizes. The experimental results demonstrate that our method exhibits a fast convergence rate and plans a path that balances path length and turn frequency while satisfying the kinematic constraints of the mobile robot. Thus, the proposed method offers an efficient and smooth path planning solution for Ackermann mobile robots in complex environments.

Numerous advanced simultaneous localization and mapping (SLAM) algorithms have been developed due to the scientific and technological advancements. However, their practical applicability in complex real-world scenarios is severely limited by the assumption that objects are stationary. Improving the accuracy and robustness of SLAM algorithms in dynamic environments is therefore of paramount importance. A significant amount of research has been conducted on SLAM in dynamic environments using semantic segmentation or object detection, but a major drawback of these approaches is that they may eliminate static feature points if the movable objects are static, or use dynamic feature points if the static objects are moved. This paper proposed DynaTM-SLAM, a robust semantic visual SLAM algorithm, designed for dynamic environments. DynaTM-SLAM combines object detection and template matching techniques with a sliding window to quickly and efficiently filter out the real dynamic feature points, drastically minimizing the impact of dynamic objects. Our approach uses object detection instead of time-consuming semantic segmentation to detect dynamic objects. In addition, an object database is built online and the camera poses, map points, and objects are jointly optimized by implementing semantic constraints on the static objects. This approach fully exploits the positive effect of the semantic information of static objects and refines the accuracy of ego-motion estimation in dynamic environments. Experiments were carried out on the TUM RGBD dataset, and the results demonstrate a significant improvement in performance in dynamic scenes.

The development of artificial microscopic robots, like synthetic microswimmers, is one of the state-of-the-art research topics due to their promising biomedical applications. The movement of microswimmers is affected by stringent constraints because of the low Reynolds number of the surrounding environment. Researchers have been working on enhancing microrobots’ propulsion tactics and figuring out new approaches to find optimal propulsion strategies. In this research, we employ a Reinforcement Learning (RL) algorithm, specifically Q-Learning, to train linear-shaped microrobots, comprised of spheres and rods, by introducing an innovative and pioneer coding approach, termed “Basic Coding”, which is utilized to specify states and actions within the RL framework. Basic Coding is a powerful general method that can be employed for different agents in any discrete RL environment. We show how to apply Basic Coding to various microrobots with different geometrical configurations, like a triangular one. Our smart microswimmers, with different numbers of spheres, acquire the knowledge of the optimal propulsion cycle to accomplish large net mechanical displacement without relying on any pre-existing locomotion expertise. The three-sphere linear microrobot recovers the cycle Najafi and Golestanian suggested. The N-sphere microrobots with higher degrees of freedom can find the optimal propulsion cycle within a reasonable number of learning steps and low computational cost utilizing our RL and Basic Coding approach, while the learning step number significantly increases using other methods like Brute-force search. For example, we show this number for the 5-sphere microrobot is 1.19E03 and 8.97E10 steps using our methodology and Brute-force, respectively. Furthermore, our intelligent microrobots can successfully and adaptively find new optimal strategies in indeterministic environments in the presence of uncertainty. Moreover, the effects of learning parameters on our RL agents are investigated in this work.

This paper proposes rigid-body modelling and identification procedures for long-reach dual-arm manipulators in a cable-suspended pendulum configuration. The proposed model relies on a virtually constrained open kinematic chain and lends itself to be simulated through the most commonly used robotic simulators without explicitly account for the cables constraints and flexibility. Moreover, a dynamic parameters identification procedure is devised to improve the simulation model fidelity and reduce the sim-to-real gap for controllers deployment. We show the capability of our model to handle different cable configurations and suspension mechanisms by customising it for two representative cable-suspended dual-arm manipulation systems: the LiCAS arms suspended by a drone and the CRANEbot system, featuring two Pilz arms suspended by a crane. The identified dynamic models are validated by comparing their evolution with data acquired from the real systems showing a high (between 91.3% to 99.4%) correlation of the response signals. In a comparison performed with baseline pendulum models, our model increases the simulation accuracy from 64.4% to 85.9%. The simulation environment and the related controllers are released as open-source code.