The article presents the development of a technical vision algorithm for automatic fruit harvesting based on the use of the YOLOv8 neural network and data obtained from binocular cameras. The algorithm uses a two-stag...
详细信息
ISBN:
(数字)9798331512194
ISBN:
(纸本)9798331512200
The article presents the development of a technical vision algorithm for automatic fruit harvesting based on the use of the YOLOv8 neural network and data obtained from binocular cameras. The algorithm uses a two-stage approach: at the first stage, the front-facing RGB-D camera installed on the mobile platform determines the position of trees and fruits, and at the second stage, the camera on the capture device specifies the coordinates of fruits using segmentation and Hough transform methods. To improve speech recognition accuracy, we use HSVspace filtering and combine data from multiple cameras using the Kalman filter. The results of modeling and experimental studies confirming the resistance to environmental influences, such as partial overlap of objects, changing lighting and the presence of foreign elements, are presented.
To address the issue of large network computing parameters for spatial-temporal features of actions in multi-view video sequences, this paper proposes a Spatial-Temporal Hypergraph Neural Network based on Attention Me...
详细信息
In this paper, we present a robust DCM-based online trajectory generator with step timing adaptation using MPC in addition to manipulating the ground reaction forces by the DCM tracking controller. The proposed contro...
In this paper, we present a robust DCM-based online trajectory generator with step timing adaptation using MPC in addition to manipulating the ground reaction forces by the DCM tracking controller. The proposed control framework utilizes three strategies to react to disturbances: timing adaptation, footstep position adjustment, and CoP modulation. Most state-of-the-art walking controllers only address some of these aspects, and especially the timing adaptation is often neglected in the presence of double support phases as the resulting optimization problem generally becomes nonlinear. We show that we can keep the fast disturbance rejection from the DCM tracking controller while adjusting the timing and location of the footsteps via MPC if the CoP-based ankle strategy is insufficient to maintain balance. This framework is particularly relevant for robots with active control of the CoP inside the support polygon by utilizing a combination of ankle and step strategy to take full advantage of the robot's capabilities in response to external disturbances. The method is validated in simulation on the robot kangaroo. It has lightweight, fully actuated legs and a sufficiently large contact area, making it highly suitable for a combined step time adaptation and contact force modulation approach.
This paper addresses the control design problem for a nonlinear and nonholonomic dynamics model of an autonomous underwater vehicle. The kinematics control algorithm is a PID controller while a self-tuning control met...
详细信息
ISBN:
(纸本)9781665454537
This paper addresses the control design problem for a nonlinear and nonholonomic dynamics model of an autonomous underwater vehicle. The kinematics control algorithm is a PID controller while a self-tuning control method is used for trajectory tracking. The kinetics control algorithm is based on an integral sliding mode technique that is robust against time-varying disturbances. The obtained results presented here use the Remote Environmental Monitoring UnitS (REMUS) underwater vehicle model and show the high performance of the controller. Based on the results demonstrated for the two designated trajectories, the tracking quality of self-tuning robust control, despite the significant similarity of the control input of both algorithms for x, y, and z, is much better than the sliding mode control. As a result, AUV was able to follow the desired trajectory, and tracking errors converged to zero. It is worth noting that at non-derivative points of the path (1 st trajectory), tracking errors were entirely within response quality limits (y = ±0.05[m]). The order of control efforts of the proposed control is suitable and can be implemented in the physical system.
Performing precise, repetitive motions is essential in many robotic and automation systems. Iterative learning control (ILC) allows determining the necessary control command by using a very rough system model to speed...
Performing precise, repetitive motions is essential in many robotic and automation systems. Iterative learning control (ILC) allows determining the necessary control command by using a very rough system model to speed up the process. Functional iterative learning control is a novel technique that promises to solve several limitations of classic ILC. It operates by merging the input space into a large functional space, resulting in an over-determined control task in the iteration domain. In this way, it can deal with systems having more outputs than inputs and accelerate the learning process without resorting to model discretizations. However, the framework lacks so far a validation in experiments. This paper aims to provide such experimental validation in the context of robotics. To this end, we designed and built a one-link flexible arm that is actuated by a stepper motor, which makes the development of an accurate model more challenging and the validation closer to the industrial practice. We provide multiple experimental results across several conditions, proving the feasibility of the method in practice.
The Rayleigh-Bénard physical model generates well-defined optical turbulence. We demonstrate its characterization and use in various experimental studies, including phase retrieval, structured light, and optical ...
The results of experimental studies of the applicability of the method of optical laser triangulation for constructing the profile of conical and cylindrical products with a ribbed surface, made by milling, are reflec...
The results of experimental studies of the applicability of the method of optical laser triangulation for constructing the profile of conical and cylindrical products with a ribbed surface, made by milling, are reflected. An algorithm for constructing a geometric profile of products of complex shape based on a system of triangulation sensors is proposed. The results of experimental approbation of the proposed algorithm on the stand for automated control of the geometry of complex-profile parts using a laser triangulation 2-D scanner are presented. The sensitivity of a laser triangulation 2-D scanner to the roughness parameters during milling the surface of products is evaluated. The accuracy characteristics of the triangulation method for determining the surface roughness parameters for different orientations of laser radiation with respect to the test object are estimated.
We compare the propagation of a Gaussian beam to the beams that carry OAM in moderate and strong underwater optical turbulence created by Rayleigh Bénard convection. Analysis focuses on scintillation index and fr...
详细信息
Human-Robot Collaboration (HRC) is people and robots work together in the same workspace to achieve shared goals. To ensure safety, robots need to perceive the position of the human body for planning purposes. However...
Human-Robot Collaboration (HRC) is people and robots work together in the same workspace to achieve shared goals. To ensure safety, robots need to perceive the position of the human body for planning purposes. However, there is no universal method for converting the human body into collision objects. To solve this problem, we propose an interface to convert the human body segments into collision objects in the robot planning scene. The interface estimates the human body Skinned Multi-Person Linear model (SMPL) by RGB image and measures the collision object size of the corresponding body segment. The corresponding size of collision objects are added to the corresponding positions of the human pose estimated by different types of sensors. The accuracy of human size estimation is verified on TotalMocap of AMASS. The interface with OctoMap is compared to evaluate the efficiency and the validity on interface is verified on the robot UR5e.
This paper investigates a dexterous manipulation method for multi-fingered hands based on the approximate policy optimization (PPO) algorithm. First, a prior knowledge base of dexterous manipulation demonstration data...
This paper investigates a dexterous manipulation method for multi-fingered hands based on the approximate policy optimization (PPO) algorithm. First, a prior knowledge base of dexterous manipulation demonstration data is provided for the PPO algorithm, and the original model is pre-trained to guide fast network training. Second, a new reward function is proposed to optimize rotational manipulation action decisions based on hand joints, fingertip tactile, and object pose infor mation. Finally, a three-fingered hand is used as an example for rotational manipulation experiments, and the comparative simulation results show that the method is more efficient than the traditional PPO algorithm and provides ideas for dexterous manipulation of multi-fingered hands.
暂无评论